north pacific research board project final reportdepts.washington.edu/fish437/resources/week...

1

NORTH PACIFIC RESEARCH BOARD PROJECT FINAL REPORT

Walleye Pollock Recruitment and Stock Structure in the Gulf of Alaska: An investigation

using a suite of biophysical and individual-based models

NPRB Project 523

Sarah Hinckley1, Carolina Parada2, John Horne3, Albert J. Hermann2, Bernard A. Megrey1,

Martin Dorn1

1National Oceanic & Atmospheric Administration, National Marine Fisheries Service,

Alaska Fisheries Science Center, 7600 Sand Point Way NE, Seattle, WA 98115. (206-526-

4109, [email protected].

2Joint Institute for the Study of Atmosphere and Oceans, University of Washington, Seattle,

WA 98195.

3School of Aquatic and Fisheries Science, University of Washington, Seattle, WA 98195.

October, 2008

Note: Mention of trade names is for information purposes only and does not constitute

endorsement by the US Department of Commerce. This report does not constitute a publication

and is for information only. All data herein are to be considered provisional and shall not be cited

without the author’s permission.

2

ABSTRACT

We developed and used a coupled model application to simulate the physical

environment and the early life history of walleye pollock (Theragra chalcogramma) in the Gulf

of Alaska (GOA). The overall goal of the modeling work has been to understand processes that

influence walleye pollock recruitment, and how recruitment may fluctuate as climate changes.

As part of this main goal, we have tried to illuminate aspects of the population structure of

walleye pollock in the North Pacific by providing a picture of relationships between spawning

locations and nursery areas. GOA and Bering Sea (BS) walleye pollock are presently managed as

a separate populations – is this management scheme justified? Spawning pollock are now found

in several different locations in the GOA, especially since the historical population spawning in

Shelikof Strait has declined. Where do the fish from these different spawning locations go? Part

of our ultimate goals with this work is to use our biophysical model to derive a model-based

index of recruitment to aid managers of walleye pollock. In order to do this, we must answer

these questions about stock structure and connectivity. In this project, we adapted and developed

a simulation modeling application to examine some of these questions, and to aid in the

development of a potential recruitment forecasting index for GOA walleye pollock. This

application consists of a coupled model set: a three dimensional model of the physical

environment, including currents, salinities, and temperatures, and an individual-based model

(IBM) of mechanisms affecting growth and survival of young walleye pollock as they move

through the environment. We present the modeling application we developed, validation

exercises, and the results of simulation experiments that shed light on the complex relationship

between spawning areas in the GOA, and the different nursery areas. This work has resulted in

increased insight into possible connections of walleye pollock within the GOA and between the

GOA and the BS.

Keywords: Walleye pollock, Theragra chalcogramma, Gulf of Alaska, Recruitment, Individual-

based Modeling, Early Life History, ROMS, Connectivity

Citation:

Hinckley, S., C. Parada, J. Horne, A.J. Hermann, B.A. Megrey, M. Dorn. 2008. Walleye pollock

recruitment and stock structure in the Gulf of Alaska: An investigation using a suite of

biophysical and individual-based models. North Pacific Research Board Final Report 523, ??? p.

3

TABLE OF CONTENTS

INTRODUCTION

OVERALL OBJECTIVES

CHAPTER 1. Java-based applications and model development for the simulation of early life history of walleye pollock in the Gulf of Alaska 1.1. Java application implementation and coupling the IBM to the hydrodynamic model 1.2. IBM model code translation and adaptation to the Java interface 1.3. Addition of new features to the biological compartments of the IBM

1.3.1. Initial conditions 1.3.2. Superindividual module 1.3.3. Water density estimation 1.3.4. Predation on juveniles 1.3.5. Active swimming of juveniles

1.4. ROMS model output preparation 1.5. Model experiment: A year of anomalous transport 1.6. Initial spawning-nursery areas experiment and parameter testing 1.7. Addition of bioenergetic and feeding algorithms for juvenile pollock to the IBM 1.8. Conclusions 1.9. Literature Cited

CHAPTER 2: Empirical corroboration of IBM predicted walleye pollock (Theragra chalcogramma) spawning-nursery area transport and survival in the Gulf of Alaska

2.1. Introduction 2.2. Methods

2.2.1. Hydrodynamic model 2.2.2. Walleye pollock IBM 2.2.2.1. Growth and bioenergetic submodels 2.2.2.2. Mortality submodel 2.2.2.3. Movement submodels 2.2.3. Model simulations 2.2.3.1. Spatial and temporal matching of model predictions and data 2.2.3.2. Prediction of potential nursery areas 2.3. Results

2.3.1. Comparison between observed and predicted distributions 2.3.2. General patterns of juvenile pollock nursery areas

2.4. Discussion 2.5. Literature Cited 2.6. Tables 2.7. Figures

4

CHAPTER 3. Connectivity of walleye pollock (Theragra chalcogramma) spawning and nursery areas in the Gulf of Alaska. 3.1. Introduction 3.2. Methods 3.2.1. Model configuration 3.2.2. Individual-based model parameters and mechanisms 3.2.3. Spatial and temporal variability of spawning areas 3.2.4. Spatial and temporal variability of nursery areas 3.2.5. Transport to and retention in nursery areas 3.3. Results 3.3.1. Retention in the spawning areas and temporal variability 3.3.2. Temporal variability of transport to nursery areas 3.3.3. Spawning in February 3.3.4. Spawning in March 3.3.5. Spawning in April 3.3.6. Spawning in May 3.4. Summary of Results 3.5. Discussion 3.6. Literature Cited

CONCLUSIONS

PUBLICATIONS

OUTREACH

ACKNOWLEDGEMENTS

LITERATURE CITED

PROJECT SYNOPSIS

5

Study Chronology

July, 2005: Project Start

January, 2006: Progress Report

July, 2006: Progress Report


July, 2007: Progress Report


October, 2008: Final Report

This project did not build directly on previously funded NPRB projects, but benefited from the expertise of many individuals supported by NPRB over the years. This project has served as foundation for other NPRB projects. We also build on projects funded by NOAA NMFS, GLOBEC and NSF support.

6

INTRODUCTION

Walleye pollock (Theragra chalcogramma) is an important commercial species in

Alaskan waters. Pollock show high interannual variability in recruitment (Dorn et al. 2005),

which is likely due to the combined effects of physical and biological processes during early life

stages (Megrey et al, 1995). Knowledge of the mechanisms behind this recruitment variability

will aid in the development of ways to forecast recruitment several years ahead of the entry of a

pollock year-class into the fishery, providing useful information for resource managers. In

addition, understanding these mechanisms should allow us to understand and predict how shifts in

climate may affect recruitment.

To understand recruitment processes for walleye pollock (Theragra chalcogramma) in

the Gulf of Alaska (GOA), and to understand how recruitment may fluctuate as climate changes,

a clearer understanding of the relationships between populations spawning in different spawning

locations and the juvenile nursery areas is needed. GOA and eastern Bering Sea walleye pollock

are presently managed as separate populations. Is this management scheme justified? Is there

transportation of eggs, larvae or juveniles between these management areas? It is necessary to

answer these questions before we can use information derived from spawning stock biomasses

and surveys of early life stages, along with mechanistic models of the success of early stages, to

develop a biophysical model-based index of recruitment.

In this project, we developed a simulation modeling application to examine some of these

questions, and to aid in the development of a potential recruitment forecasting index for GOA

walleye pollock. This application consists of two previously existing models linked to form a

coupled model set: a three dimensional model of the physical environment, including currents,

salinities, and temperatures, and an individual-based model (IBM) of mechanisms affecting

growth and survival of young walleye pollock as they move through the environment. Here we

present the developments of the coupled model set (Parada et al., accepted) and the interface to

these models that we developed. We also present model corroboration exercises and the results

of simulation experiments that shed light on the complex relationship between spawning and

nursery areas in the GOA and the BS.

Two ultimate goals of this work were: (1) to explore whether our coupled biophysical

model can be used to generate an index of recruitment several years ahead of its entry into the

7

fishery, and (2) to understand the biophysical mechanisms behind the variability in recruitment,

which will allow us to explore future effects of climate change on recruitment. The application

that we have developed will not only be useful for the study of walleye pollock recruitment in the

GOA, but also to study recruitment of pollock and other species in other areas, such as the Bering

Sea and the West Coast of North America. It is a versatile and powerful tool.

In Chapter 1, we describe additions to the physics and to the biology of the walleye

pollock IBM precursor model, the application that we have developed and the validation tests

accomplished before the model was deemed ready for use. Details of the precursor models can

be found in Haidvogel et al., 2000; Hermann et al. 1996a; Hermann et al. 1996b; Hermann &

Stabeno 1996; Hermann et al. 2001; Hinckley et al. 1996; Hinckley et al. 2001; Megrey &

Hinckley, 2001; Moore et al., 2004; and Shchepetkin and McWilliams, 2004. In Chapter 2, we

describe model corroboration studies, which found the model capable of correctly simulating the

movement through space of pollock eggs, larvae and juveniles, and of simulating the known

nursery areas of GOA walleye pollock. In Chapter 3, we describe our final study for this project

exploring the relationships or connectivity between GOA pollock spawning areas and nursery

areas. This information will aid managers in making informed decisions about how to effectively

manage this species, and also to understand the best way to use the coupled model set as a

forecasting tool. This work will indicating what indices the model may be able to produce, and

what measures may be well correlated with recruitment.

OVERALL OBJECTIVES

In this project, we proposed to refine and further develop a suite of coupled models and

to use these to investigate recruitment variation in, and stock structure of walleye pollock in the

Gulf of Alaska. We originally proposed to develop two outputs: (1) an index of abundance of

pre-recruits for managers of this stock, and (2) information on stock structure of pollock in the

Gulf, as determined by the modeled success of spawning and retention of juveniles in different

regions within the Gulf. We utilized a hydrodynamic model (Regional Ocean Modeling System,

ROMS), adapted to this region, and we further developed our individual-based model (IBM) of

the early life of walleye pollock to meet these objectives. This project directly addresses NPRB’s

mandate to fund research which furthers our understanding of, and ability to manage, fish

populations in the Gulf of Alaska region of the North Pacific.

8

The objectives of this project were, of necessity, modified as we progressed through this

work. The model and coupled model application took significantly more time than anticipated to

develop. We decided to focus on the stock structure aspect of the work (rather than the

recruitment prediction), since without better understanding of stock structure and connectivity of

walleye pollock spawning and nursery areas in the GOA, we would not be in a position to derive

a pre-recruitment index from this coupled model set. This goal was accomplished.

The overall objectives, as modified, were to:

1. Adapt and develop a Java application that allows us to use physical model output from the

ROMS models, in conjunction with the walleye pollock IBM, to simulate the early life history of

walleye pollock in the Gulf of Alaska in a user friendly, flexible manner. This application is

easily adaptable for use in different simulation configurations and even different species and

areas.

2. Make needed improvements to the walleye pollock IBM within the Java application, to reflect

new thinking about factors governing connectivity and recruitment variability.

3. Perform model simulation experiments designed to test and corroborate the coupled model set.

4. Perform model simulation experiments designed to illuminate the relationships between

different spawning areas and nursery areas of walleye pollock in the GOA. This information will

be used to aid managers in assessing management strategies, and in designing potential indices to

be derived from the coupled model in future studies which may be useful in forecasting year class

strength.

9

CHAPTER 1: Java-based application and model development for the simulation of early life history of walleye pollock in the Gulf of Alaska 1.1 Java application implementation and coupling the IBM to the hydrodynamic model

A Java programming application was adapted to simulate the early life history of walleye

pollock in the Gulf of Alaska (GOA). The purpose of the study was to understand the transport

and survival of walleye pollock from spawning to nursery areas. This application enabled us to

track trajectories of simulated particles representing individual or groups of individual fish, and

water properties (temperature, salinity) experienced by particles during transport. It uses three-

dimensional fields of velocity, temperature, and salinity archived from simulations of the

Regional Oceanic Modeling System (ROMS, Haidvogel et al., 2000; Moore et al., 2004;

Shchepetkin and McWilliams, 2004) to track particle trajectories within the model domain. As

described in Lett et al. (2008), the application offers two functioning modes. The first allows a

visualization of the transport of simulated eggs and larvae in a user-friendly graphic interface

showing the trajectories of all particles being transported, and the evolution of selected variables

followed during simulations. Examples of variables that can be tracked include distributions of

length or weight, growth curves, or number of survivors. The second mode enables batch

simulations based on pre-defined sets of parameters and produces output files for each series. The

program stores information on the dynamics of individuals (e.g. time, longitude, latitude, depth,

length, etc.). Output files are in ASCII format for ease of post-processing. The application is

distributed as a package that contains the program code and libraries (for details see

http://www.ur097.ird.fr/projects/ichthyop/). For this project, we used one of the early releases of

the Java application and adapted it to our model requirements (see below).

The Java application was coupled to the ROMS hydrodynamic model to simulate the

transport of particles released in spawning areas for walleye pollock in the GOA (Figure 1.1). We

used a subset of the output of the multi-decadal coupled sea-ice/ocean numerical simulations of

the Northeast Pacific (NEP) region configured and run by Curchitser et al. (2005). The coupled

model is based on the ROMS model implemented at 10 km resolution for a Northeast Pacific

domain, which includes the Gulf of Alaska and the Bering Sea (Figure 1.2).

10

Figure 1.1. Java graphical interface showing SST and currents from ROMS model output, and in red initial positions of drifters. To the left are slider bars where parameters can be set. At the bottom are histograms or graphs showing model output.

Figure 1.2. Nested grids used in the ROMS model system. The NPac grid, outlined in red, has 20-40 km horizontal resolution. The NEP grid, outlined in light green, has 10 km horizontal resolution. The CGOA grid, the northernmost of the two grids outlined in light blue, has a horizontal resolution of 3 km. The coupled model described here was run on the NEP grid.

11

A series of experiments was conducted to test the reliability of the particle tracking

algorithm within the Java application. Two tracking algorithms were tested: one using the Euler

method for solving differential equations, the other using a 4th order Runge-Kutta method. In

these experiments, the number of iterations (i) between time steps was varied from 1 to 10,000.

The Runge-Kutta method was more stable than the Euler method, and converged quickly (i < 10)

to the (x, y) position (Fig. 1.3a and b). The Euler method took longer to converge (100 > i > 200)

(Fig. 1.3a and b).

Figure 1.3. Convergence of particle positions (a) x and (b) y after a series of simulations varying the number of iterations between time steps using two algorithms for particle tracking (Euler and Runge-Kutta).

The error in displacement before convergence was very high for a low number of

iterations when using the Euler method (Fig. 1.4). For i < 100, the error in displacement was

higher (~130 km) than the spatial resolution of the ROMS model output. The error in

displacement when using Runge-Kutta

method, was very low (Fig. 1.4).

Figure 1.4. Displacement of particles from the final position after ‘i’ iterations using two algorithms for tracking particles: Euler and Runge-Kutta.

12

1.2. IBM model code translation and adaptation to the Java interface

The original IBM model for walleye pollock was coded in the C programming language

(Hinckley et al., 1996; Megrey and Hinckley, 2001). It has now been translated into Java, and a

new model interface has been completed. All biological compartments and subroutines were

translated from C to Java.

1.3. Addition of new features to the biological compartments of the IBM

(See Chapter 2 for details of algorithms and parameters mentioned in the next sections)

1.3.1. Initial Conditions

Algorithms were written to select spawning areas, the depth of spawning, and the timing

of spawning. Eggs can be released randomly or with a stratified random distribution, at specific

locations, depths, and times.

1.3.2. Superindividual module

Superindividual schemes (Scheffer et al., 1995; Megrey and Hinckley, 2001; Bartsch and

Coombs, 2004) are approaches that allow the use of realistic mortality rates in IBMs by

increasing the number of individuals represented by each point or float (i.e. superindividual). In

this way, we can simulate larger numbers of fish in a population without significantly adding to

computer processing time. The assumption behind this approach is that growth, feeding

conditions, and the probability of mortality for each individual within a superindividual are the

same. At the beginning of the simulation, each superindividual is assigned a “count”, indicating

how many individual fish the superindividual represents. This “count” is decreased by applying a

random deviate from the daily mortality rate for that particular life stage. At any point in time, the

total number of individual fish is the sum of the “count” variable over all superindividuals. The

value of the “count” of the superindividual at time t+1 depends on that of the superindividual at

time t, the mortality rate r, and the time step t.

1rt

t tS S e−+ = (1)

13

Mean = 0, St dev. Turning angle = 5



a

b

c

1.3.3. Water density estimation

The UNESCO equation for the estimation of seawater density based on salinity and

temperature was incorporated in the code to replace a linear function of salinity. The original

algorithm was intended for a model which covered a more restricted domain (Parada et al.,

accepted) for which the linear function of density with salinity was appropriate. However the

larger domain covered by the ROMS model necessitated the use of the more general UNESCO

function.

1.3.4. Predation on Juveniles

Groundfish predation on 0-age juveniles was added to the juvenile subroutine. Predation

rates were based on 8 years of groundfish predation data (K. Aydin, Alaska Fisheries Science

Center, Seattle, WA, pers. comm.). For years where no data were available, an average rate over

the 8 years was used. The predation rate was calculated using the total numbers of walleye

pollock juveniles consumed by all groundfish predators in each year. We computed mortality

rates assuming that the maximum number consumed was equivalent to a predation mortality rate

of 0.3 per day (average). Predation rates for other years were standardized relative to this

maximum.

1.3.5. Active

swimming of juveniles

A correlated

random walk

algorithm was added

to the model to

simulate horizontal

movement of

juveniles. Examples

of trajectories of a

single individual (60

mm length) with

mean turning angle of

14

0o and several different values for the standard deviation of the turning angles are shown in

Figure 1.5.

Figure 1.5. Trajectories of juvenile pollock at 60 mm using different values for the standard deviation of the turning angle.

A standard turning angle deviation of 20 degrees was used in the simulations described

below. When this value was used, the overall distribution of 0-age juveniles was closer to the

Alaska Peninsula and patchier than simulations without active swimming of juveniles, which was

what we were trying to accomplish by this addition to the model, as it is what the observations

indicate.

1.4. ROMS model output preparation

Outputs from ROM model runs were corrected for the time series 1978-2004 to fit the

requirements of the IBM, and some faulty files were replaced. We are currently using a subset of

this time series (several years in the 1980’s, the 1990’s, and the years 2000-2004) to perform

model experiments examining walleye pollock spawning-nursery area connections.

1.5. Model experiment: A year of anomalous transport

The model was run for 1985, a year for which we have data showing anomalous transport

patterns of pollock eggs associated with changes in salinity (K. Bailey, Alaska Fisheries Science

Center, pers. comm.). The model output matched patterns contained in the empirical data:

transport of eggs in a northeasterly direction through the Shelikof Strait. Pollock eggs are usually

transported to the southwest through Shelikof Strait.

Figure 1.6. Locations of egg release in initial spawning-nursery area experiment. Areas are located around Kodiak Island.

1.6. Initial spawning-nursery areas experiment and

parameter testing

15

A second series of simulations was performed to study the relationship between spawning

and nursery areas for the years 2000-2004. Eggs were randomly released in six areas in Shelikof

Strait and around Kodiak Island (Fig. 1.6). The eggs were released at an average depth of 200 m,

or at the bottom when depth was shallower than 200 m. The period of egg release was between

1st March and 1st April assuming a normal distribution.

Five years of ROMS output were used to provide the physical flow fields in the IBM for

this experiment. A dynamic prey source was not used. A constant consumption function based on

weight was used to regulate development of larvae and juvenile fish. The number of individual

floats released each year was 5000. We also used the superindividual approach in the model,

where each superindividual modeled represented a cohort of fish (as above) with the number

proportional to the egg production for that year (given by stock assessment model values). These

superindividuals were subject to a daily natural mortality caused by predation. The simulation

duration was 3 months (i.e. early March until early June). At the end of the simulation, the

trajectory, destination, stage (feeding larvae or juvenile), size, and count (representing the number

that survived for each superindividual) were tabulated.

Figure 1.7 (right). Final length frequency histograms for annual simulations. The 5 years of simulations were: A) 2000, B) 2001, C) 2002, D) 2003, E) 2004.

16

0

500

1000

1500

2000

2500

3000

3500

0 5 10 15 20 25 30 35 40 45 50 55 60

Size class (mm)

Freq

uenc

y

larvae 2000juvenile 2000

A

0

500

1000

1500

2000

2500

3000

3500

0 5 10 15 20 25 30 35 40 45 50 55 60

Size class (mm)

Freq

uenc

y


B

0

500

1000

1500

2000

2500

3000

3500

0 5 10 15 20 25 30 35 40 45 50 55 60

Size class (mm)

Freq

uenc

y


C

0

500

1000

1500

2000

2500

3000

3500

0 5 10 15 20 25 30 35 40 45 50 55 60

Size class (mm)

Freq

uenc

y


D

0

500

1000

1500

2000

2500

3000

3500

0 5 10 15 20 25 30 35 40 45 50 55 60

Size class (mm)

Freq

uenc

y


E

For the year 2000 (Fig. 1.7A)

slower growth and development of the

individuals was observed compared to

the other years, with lengths of larvae at

the end of the simulation between 15-25

mm. Simulation results from 2001 and

2003 showed mainly the presence of

juveniles with sizes between 30-50 mm.

The years 2002 and 2004 contained a

wide range of size classes (20-50 mm),

with both larvae and juveniles present.

The ranked order of growth was

2000<2002<2004<2003<2001.

Figure 1.8. Frequency of superindividual counts for IBM simulations for 5 years A) 2000, B) 2001, C) 2002, D) 2003, E) 2004.

Frequency distributions of the

number of fish represented by each

superindividual (superindividual counts,

Fig. 1.8) showed that during 2000 high

counts of larvae (105-107), indicated that

superindividuals experienced less

mortality than in the other years. In

years 2001, 2003 and 2004, the fish

surviving to the end of the simulation

were mainly juveniles, which had

experienced a higher cumulative

mortality (i.e. counts of only ~103

juveniles). At the end of the simulation

for year 2002, both larvae and juveniles

were present, with superindividual

17

1.0E+001.0E+011.0E+021.0E+031.0E+041.0E+051.0E+061.0E+071.0E+081.0E+091.0E+10

2000 2001 2002 2003 2004

Year

Supe

rindi

vidu

al s

ize

(Num

ber) larvae

juveniles

0

1000

2000

3000

4000

5000

1000 10000 100000 1000000 10000000

Superindividual class

Freq

uenc

y


A

0

1000

2000

3000

4000

5000

1000 10000 100000 1000000 10000000


Freq

uenc

y


B

0

1000

2000

3000

4000

5000

1000 10000 100000 1000000 10000000


Freq

uenc

y


C

0

1000

2000

3000

4000

5000

1000 10000 100000 1000000 10000000


Freq

uenc

y


D

0

1000

2000

3000

4000

5000

1000 10000 100000 1000000 10000000


Freq

uenc

y


E

counts of 103-105 (Fig. 1.8).

The total number of fish that

survived (sum of all superindividual

counts) for each stage showed that 2000

had the highest number of larvae due to

low mortality, but that no juveniles

survived due to the low growth rates (Fig.

1.9). The overall number of juveniles for

2001-2004 was very similar. Results from

2001 showed a small number of surviving

larvae, with an increasing trend toward

Figure 1.9. Total number of pollock surviving until the end of the simulations for 5 years (2000 to 2004), in larval and juvenile stages.

2004. This may mean that 2004 had the

potential to increase the overall abundance

of juveniles compared to 2001.

The overall total numbers of surviving individuals by spawning area, showed interannual

variability (Table 1.1). For larvae, areas 3 and 5 were the most successful spawning grounds

during 2000, with the highest numbers of survivors. In 2001, few or no larvae survived in any

area. In 2003 there were no larvae from area 3, and in 2004, there were no larvae observed in

area 5. For juveniles in 2001, spawning areas 0, 4, and 5 were the most successful. In 2002,

18

spawning areas 3 and 5 were successful, and in 2003, area 3 was the most successful in producing

juveniles.

Table 1.1A. . Overall total numbers by spawning areas for walleye pollock larvae. The spawning

areas are shown on Figure 1.6

Year

Area of Origin

2000 2001 2002 2003 2004

0 224,222,876 1 43,773 280,681 1,798,595

1 245,389,679 29 54,592 405,584 879,244

2 253,171,175 18 21,595 546,590 901,9513 1,187,468,054 0 658 0 134,0164 553,041,453 0 8,985 1,614,660 198,8835 1,111,716,827 0 663 352,651 0

Table 1.1B. Overall total numbers by spawning areas for walleye pollock juveniles.

Year

Area of Origin

2001 2002 2003 2004

0 3,654,715 10,724 126,933 766,683 1 2,142 28,026 285,408 226,894 2 7,848 69,259 183,166 320,126 3 425,255 1,895,301 1,020,924 196,457 4 935,084 749,964 320,562 98,120 5 1,226,132 2,495,944 236,649 18,433

This simulation was not designed to indicate recruitment strength, as it only runs until the

beginning of June, and has no dynamic prey source. However, it does tell us that there is

significant interannual variability in the number of surviving individuals from the different

spawning areas. It also tells us that other regions around Kodiak besides Shelikof Strait may be

important to the production of walleye pollock in the GOA. These are both important findings.

The next set of simulations included realistic initial conditions for selected years and a wider

range of spawning locations.

19

1.7. Addition of bioenergetic and feeding algorithms for juvenile pollock to the IBM

We added a new bioenergetics submodel to the IBM for 0-age juveniles. From the earlier

experiments, it was noted that the original version of the juvenile bioenergetic equations resulted

in higher than observed growth rates. Ciannelli et al.’s (1998) bioenergetics model performed

better relative to the data and was therefore added to our IBM. This model also contained

digestion, following Mazur et al. (2007), which was not present in the earlier version.

A feeding model was also added for juvenile pollock. This model was based on field

data showing prey preference depending on juvenile walleye pollock size (M. Wilson, AFSC,

Seattle, pers. comm), with an increasing shift to euphausiids from copepods as juvenile fish grew.

Prey availability, in the absence of the NPZ model, was constant, but proportions of each type

(small and large copepods, euphausiids) in each area were based on historical data (M. Wilson,

AFSC, Seattle, pers. comm., NPRB Project 308 Final Report). Prey was set to 0 at depths

>1000m (ie. off the continental shelf). In the inner shelf areas, the density of euphausids, small

copepods, and large copepods were higher (1.17838, 486.2773, 66.66362 number m-3). In the

mid and outer shelf and slope areas, prey densities started low (number of euphausids, small and

large copepods were 0.294595, 121.56933, 16.665905 number m-3) up to Day of the Year (DOY)

120 and then increased up to DOY 160 (mid June) to the values for the inner shelf; based on NPZ

model output for several years (Hinckley, 1999).

20

Figure 1.10. Output of the model simulation testing bioenergetics and feeding of juveniles. a. change in weight (g) between August 20-25, b. change in length (mm) between August 20-25, c. distribution of juvenile lengths on 20 August, and d. distribution of lengths on 25 August.

For the simulations to test the bioenergetics and feeding behavior rules, the model was

run for 45 days (Aug-Sept), starting with a random distribution of juvenile fish lengths. One

thousand individuals were released between Shelikof Strait and the Shumagin Islands. As noted

above, the prey distribution had a coast-ocean gradient, with copepods and euphausiids more

abundant inshore.

The results of this simulation showed individuals clustered over the outer shelf and slope,

with very few transported to the nearshore regions. Figures 1.10a and b show the change in

juvenile weight and length in the areas where the fish were found (where delta weight was zero)

a b

c d

21

from August 20 to August 25. The maximum change in weight in 5 days (Fig. 1.10a) was ~ 3.0 g

(ie. ~0.6 g d-1), which is reasonable. The spatial distribution of length is shown in Figs. 1.10c (20

Aug.) and 1.10d (25 Aug.).

A second test of the new biology incorporated the early life history of pollock (from eggs,

to early juveniles) and all of the biological mechanisms (spawning, buoyancy, vertical migration,

swimming behavior, growth, mortality, starvation). The objectives of this simulation were to: 1)

observe interannual variability in model variables for larvae and juvenile stages, 2) estimate the

density of pollock at age-0 based on superindividual features using GIS, and 3) test the

configuration to be used for the final connectivity analysis (see Chapter 3).

Figure 1.11. Source and sink areas used to build the connectivity matrix. Spawning (source) areas were 2, 3, 5, 6, 8, 9, 11, 12, 14, 15, 17, 18, 20, and 21. Sink areas were 0-46.

For this test, the model was run for the years 2000 to 2004. We released eggs randomly over a

broad time period (February-June) over the continental shelf and slope (Prince William Sound

(PWS) to the Shumagin Islands) of the GOA. Source and sink areas were defined using GIS to

establish a connectivity matrix (Fig. 1.11 and Chapter 3). We released 5000 individuals in 14

spawning areas from PWS to Unimak Pass (Fig.1.12A).

30

43

0

38

34

42

39

44

3229

35 3

28

4546

4140

636

33

79

37

1925

10

14

23

5

24 22

31

13161820

17 1215

8

2

1411

21Sink/source areas

22

Figure 1.12. A. Regions where modeled pollock eggs were released for the years 2000-2004. B. Positions of particles (corresponding to eggs, yolksac larvae, feeding larvae, and juveniles) during the simulation on 1 August, 2000.

Overall results for this simulation for the years 2000-2004 are shown in Figure 1.13 and

1.14. Larval hatch size and larval first feeding weight modeled values agreed with literature

values (Fig. 1.13A, B). Larval first feeding dates occurred in 3 distinct pulses: the first on day 20

of the simulation, between day 40 and day 90, and a third pulse at day 120 of the simulation (Fig.

1.13C). The second pulse

Figure 1.13. Simulation model output testing bioenergetics and feeding of juveniles and new model biology. A. Larval hatch size, B. larval first feeding weight, C. larval first feeding date, D. superindividual number.

30

43

0

38

34

42

39

44

3229

35 3

28

4546

4140

636

33

79

37

1925

10

14

23

5

24 22

31

13161820

17 1215

8

2

1411

21

30

43

0

38

34

42

39

44

3229

35 3

28

4546

4140

636

33

79

37

1925

10

14

23

5

24 22

31

13161820

17 1215

8

2

1411

21

Spawning region Position 1st August 2000

A B

3.75

3.85

3.95

4.05

4.15

200020012002200320040

0.1

0.2

0.3

0.4Proportion

Larval hatchsize (ml)

Year

0.3-0.40.2-0.30.1-0.20-0.1

A

0 10 20 4060

80100

1202000

2001

2002

2003

2004

00.1

0.2

0.3Proportion

Larval feeding date

Year

0.2-0.30.1-0.20-0.1

C

20 21 22 23 24 25 26 272000

20012002

200320040

0.050.1

0.150.2

0.250.3

0.35Proportion

Larval feeding weigth (ug)

Year

0.3-0.350.25-0.30.2-0.250.15-0.20.1-0.150.05-0.10-0.05

B

0.E+

00

1.E

+03

1.E

+05

1.E

+07

1.E

+09 2000

20012002

20032004

0

0.1

0.2

0.3

0.4Proportion

Superindividual number

Year

0.3-0.40.2-0.30.1-0.20-0.1

D

23

corresponded to eggs spawned on days 30 and 60 of the simulation. During 2001 we observed a

larger proportion of larvae starting to feed on day 100 (Fig. 1.13C).

The overall number of juvenile survivors was calculated by adding all juveniles that

survived through September 30th of each year, times the superindividual number. We observed

that survival was larger in 2001 compared to the other simulation years. A trend in increasing

survival occurred from 2002 to 2004 (Fig. 1.14).

0.0E+00

5.0E+07

1.0E+08

1.5E+08

2.0E+08

2.5E+08

3.0E+08

3.5E+08

2000 2001 2002 2003 2004

Year

Juve

nile

num

ber

end

of S

epte

mbe

r

Figure 1.14. Number of juveniles at the end of the simulation (30th September) each year.

1.8. Conclusions

Significant methodological advances were made to the coupled model set during this

project. We adapted a Java application to our previous pollock IBM which gives us a graphical

user interface, makes the model faster and easier to run, and allows us to use ROMS output to

provide physical flow fields. Extensive testing of the float tracking algorithm gave us confidence

that the simplest method (Euler) would be adequate, if we used a higher number of iterations.

The original C code was translated into Java with all the biological mechanisms that were

included in the original model (Hinckley et al., 1996; Megrey and Hinckley, 2001). Important

24

new features were added to the simulation model: an algorithm to set initial conditions of egg

release (spatial and temporal), a superindividual module, a new algorithm to estimate water

density, estimates of groundfish predation on 0-age juveniles based on data, and active swimming

by juveniles. We also obtained and corrected the ROMS model output for 1978-2004 to use with

these model simulations.

We performed several model simulations to test model additions and to simulate

particular phenomena, to see how well the model performed. We succeeded in simulating the

anomalous transport seen in 1985. In an initial spawning-nursery area experiment and parameter

test, we found that the model could simulate significant interannual variability in larval and

juvenile length distributions and mortality, and found differences in the success of different

spawning areas, including those outside of Shelikof Strait. This test also allowed us to review

parameters included in the model.

We did several more tests of juvenile bioenergetics, movement, and feeding modules.

The first test, focusing on the early juvenile phase, resulted in reasonable growth of juvenile

walleye pollock. The second test included the whole early life history from eggs to juveniles in

the fall. We found that larval variables matched values from the literature.

In conclusion, we now have an application which can reliably be used to test the main

hypotheses of this project about spawning and nursery area connectivity.

1.9. Literature Cited

Bartsch, J. and S. Coombs. 2004. An Individual-Based Model of the early life-history of mackerel

(Scomber scombrus) in the eastern North Atlantic, simulating transport, growth and mortality.

Fisheries Oceanography, 13: 365-379.

Ciannelli, L., R.D. Brodeur and T.W. Buckley. 1998. Development and application of a

bioenergetics model for juvenile walleye pollock. Journal of Fish Biology, 52:879-898.

Curchitser, E.N., D.B. Haidvogel, A.J. Hermann, E.L Dobbins, T.M. Powell and A. Kaplan.

2005. Multi-scale modeling of the North Pacific Ocean: Assessment and analysis of simulated

basin-scale variability (1996-2003). Journal of Geophysical Research. C. Oceans 110(C11), [np].

25

Dorn. M., Aydin. K., Barbeaux. S., Guttormsen. M., Megrey. B., Spalinger. K. and M. Wilkins.

2005. Assessment of walleye pollock in the Gulf of Alaska In: Appendix B. Stock assessment and

fishery evaluation report for the Groundfish Resources of the Gulf of Alaska, North Pacific

Fishery Management Council, P.O. Box 103136, Anchorage, AK 99510.

Haidvogel, D.B., H.G. Arango, K. Hedstrom, A.Beckmann, P. Malanotte-Rizzoli, P. and A.F.

Shchepetkin. 2000. Model evaluation experiments in the North Atlantic Basin: Simulations in

nonlinear terrain-following coordinates, Dynamics Atmosphere Oceans 32: 239-281.

Hermann, A.J., Hinckley, S., Megrey, B.A. and P.J., Stabeno. 1996a. Interannual variability of

the early life history of walleye Pollock near Shelikof Strait as inferred from a spatially explicit,

individual-based model. Fish Oceanogr 5(Suppl.1):39-57.

Hermann, A.J., Rugen, W.C., Stabeno, P.J., and N.A., Bond. 1996b. Physical 828 transport of

young pollock larvae (Theragra chalcogramma) near Shelikof Strait as inferred from a

hydrodynamic model. Fish Oceanogr 5(Suppl.1):58-70.

Hermann, A.J. and P.J., Stabeno. 1996. An eddy-resolving model of circulation on the western

Gulf of Alaska shelf. 1. Model development and sensitivity analyses. J Geophys Res 101(CI):

1129-1149.

Hermann, A.J., Hinckley, S., Megrey, B.A. and J.M., Napp. 2001. Applied and theoretical

considerations for constructing spatially explicit individual-based models of marine larval fish

that include multiple trophic levels. ICES J Mar Sci 58:1030-1041.

Hinckley, S., Hermann, A.J. and B.A. Megrey, 1996. Development of a spatially-explicit,

individual-based model of marine fish early life history. Marine Ecology Progress Series, 139:

47-68.

Hinckley, S. 1999. Biophysical mechanisms underlying the recruitment process in walleye

pollock (Theragra chalcogramma). PhD dissertation, University of Washington, Seattle. 258 pp.

26

Hinckley. S., Hermann, A.J., Mier, K.L. and B.A., Megrey. 2001. Importance of spawning

location and timing to successful transport to nursery areas: a simulation study of Gulf of Alaska

walleye pollock. ICES J Mar Sci 58:1042-1052.

Mazur, M.M., M.T. Wilson, A.B. Dougherty, A. Buchheister and D.A. Beauchamp. 2007.

Temperature and prey quality effects on growth of juvenile walleye pollock Theragra

chalcogramma (Pallas): a spatially explicit bioenergetic approach. Journal Fish Biology, 70: 816-

836.

Megrey, B.A., Bograd, S.J., Rugen, W.C., Hollowed, A.B., Stabeno, P.J., Macklin, S.A.,

Shumacher, J.D. and W.J. Ingraham. 1995. An exploratory analysis of associations between

biotic and abiotic factors and year-class strength of Gulf of Alaska walleye pollock (Theragra

chalcogramma). In: Beamish RJ (ed) Climate change and northern fish populations. Can Spec

Pub Fish Aquat Sci 121:227-243.

Megrey, B.A. and S. Hinckley. 2001. Effect of turbulence on feeding of larval fishes: a

sensitivity analysis using an individual-based model. ICES Journal of Marine Science 58: 1015-

1029.

Moore, J. K., D.M. Doney, C. Scott, and K. Lindsay, 2004. Upper ocean ecosystem dynamics

and iron cycling in a global three-dimensional model. Global Biogeochemical Cycles 18, [np].

Scheffer, M., J.M. Baveco, D.L. DeAngelis, K.A. Rose and E.H. Van Nes. 1995. Super-

individuals: a simple solution for modelling large populations on an individual basis. Ecological

Modeling, 80: 161–170.

Shchepetkin, A.F. and J.C. McWilliams. 2005. The Regional Ocean Modeling System: A split-

explicit, free-surface, topography-following coordinate ocean model, Ocean Modeling, 9: 347-

404.

27

CHAPTER 2: Empirical corroboration of IBM-predicted walleye pollock (Theragra chalcogramma) spawning-nursery area transport and survival in the Gulf of Alaska (Draft Manuscript – Citation is not allowed without author permission) C. Paradaa,c, S. Hinckleyb, J. Hornec , M. Mazura

aJoint Institute for the study of Atmosphere and Ocean (JISAO) University of Washington, Box 355672, Seattle, WA 98195 bAlaska Fisheries Science Center, National Oceanographic and Atmospheric Administration (NOAA), 7600 Sand Point way NE, Seattle, WA, 98115 cSchool of Aquatic and Fishery Sciences, University of Washington, Box 355020, Seattle, WA, 98195 2.1. Introduction

Understanding the role of nursery areas in the recruitment of marine fish populations enables

fisheries managers to target conservation efforts and improve management decisions that preserve

diversity and protect natural resources. Even though the term “nursery area” is broadly used in

the ecological literature, there is not a consistent definition. The term nursery area was first

applied by Harden Jones (1968) to describe the movement by fish species with complex life

cycles and larval stages that are transported to estuarine systems where they grow and then move

to adult habitats (Deegan, 1993). Beck et al. (2001) reviewed the role of nursery areas for marine

fish and invertebrates and defined the term as an area where the contribution per unit area to the

production of recruits to the adult populations is greater, on average, than production from other

regions in which juveniles occur. In their definition, nursery habitats must support an important

contribution to adult recruitment which result from any combination of factors such as density,

growth, survival of juveniles, and movement to adult habitat (Beck et al., 2001). We assume

herein that nursery areas are habitats characterized by high growth, density, and survival of

juveniles.

Walleye pollock is an important fishery in the Gulf of Alaska (GOA), and has a life history

where spawning and nursery habitat are separated. Currents in the region where pollock spawn

are advective, but information on the location and relative importance of spawning and nursery

areas in the GOA is limited, as surveys do not cover all spawning and nursery areas every year.

Historically, a majority of walleye pollock spawning was located between Kodiak Island and the

Alaska Peninsula in an area known as Shelikof Strait (Fig. 2.1; Kendall et al., 1987; Schumacher

and Kendall, 1991). Positively buoyant eggs are released between mid-March and early May in

this area, with peak spawning at the beginning of April. By May, early larvae are advected

southwest in the Alaska Coastal Current (ACC) along Alaska Peninsula. By summer and through

28

early fall, juveniles arrive at the Shumagin Islands nursery area (Hinckley et al., 1991; Spring and

Bailey 1991; Wilson et al., 1996; Hinckley et al., 2001). The Shumagin Islands region was

proposed to be the nursery area that ensures success of the GOA walleye pollock population

(Hinckley et al., 1991; Wilson et al., 1996). More recently, other potential spawning and nursery

areas have been reported (Bailey et al., 1999; Wilson, 2000; Mazur et al., 2007), but their relative

contribution to the GOA population is not known. Ciannelli et al (2007) discuss the relative

importance and stability of spawning areas over time. Spatial and temporal variability in

spawning and variability in current flows may determine arrival at and use of nursery habitat in

GOA and explain some of the variability in walleye pollock recruitment.

A suite of biophysical models consisting of an individual-based model (IBM) coupled to a

hydrodynamic model, can be used to assess which GOA areas have the potential to be nursery

areas for juvenile walleye pollock. The spatially-explicit and Lagrangian features of biophysical

models allow tracking of individuals from release locations to destinations and recording of

trajectories during this period. Previous efforts using this technique in the GOA were used to

explore the life history of walleye pollock (e.g. Hinckley et al., 1996, Hinckley et al., 2001,

Megrey & Hinckley, 2001; Hermann et al., 2001) and to examine how biological and physical

factors influence recruitment (Parada et al., accepted).

The utility of IBM models to assess the potential connectivity between spawning andnursery

areas has not been determined. Can this set of coupled models reproduce distributions of different

life stages of walleye pollock, including the juvenile nursery areas, observed during a year with

surveys of spawning adults, larvae, and September 0-age juveniles? If so, then we should be able

to use the model to examine connections between spawning and nursery areas. If we randomly

scatter pollock eggs over the known range of pollock spawning in the GOA, and tabulate

locations and abundances of juveniles at the end of the simulation, do distribution and abundance

patterns resemble what we have observed? This study examines spatial and temporal coincidence

between model outputs and observed distributions of walleye pollock successive life stages in the

GOA during a single year (1987), and analyzes modeled nursery areas as compared to observed

ones in the GOA.

29

2.2. Methods

2.2.1 Hydrodynamic model

The biophysical model couples the Regional Ocean Modeling System (ROMS)

circulation model with an individual based model (IBM) of walleye pollock life history. ROMS

is a free-surface, hydrostatic primitive equation, ocean circulation model that employs a nonlinear

stretched vertical coordinate to follow the bathymetry. Horizontal space is discretized using

orthogonal curvilinear coordinates on an Arakawa C grid. Numerical details can be found in

Haidvogel et al. (2000), Moore et al. (2004), and Shchepetkin and McWilliams (2005). Within

the ROMS model, water current and temperature resolution in the Gulf of Alaska (GOA) was set

at 10 km using the Northeast Pacific (NEP) grid (Fig. 1.2, see Curchitser et al. (2005) for physical

model details). First, 1987 was simulated, for the purposes of model comparison to data. Then

six more years were simulated to examine whether the model was capable of replicating observed

juvenile nursery areas in an average sense (ie. over more than year) using 3 years prior to (1978,

1982, 1988) and 3 years after (1992, 1999, 2001) the shift in the control of recruitment in the

GOA from Bailey’s hypothesis (Bailey, 2000). The ROMS simulations were run with forcing

(winds, freshwater runoff, and boundary conditions) appropriate for each year. The

hydrodynamic model produces daily averaged output consisting of salinity and temperature

fields, and 3D water velocities, which were used to drive the IBM.

2.2.2. Walleye pollock IBM

The walleye pollock IBM simulated four development stages (eggs, yolk-sac larvae,

feeding larvae, and juveniles) from spawning to the fall of the 0-age year. The IBM of Hinckley

et al. (1996) and Megrey and Hinckley (2001) was expanded to include bioenergetic and

swimming submodels for juveniles plus coupling to the ROMS model instead of the Spectral

Primitive Equation Model (SPEM).

2.2.2.1. Growth and bioenergetic submodels

Mortality and growth were stage-dependent in the IBM and were modeled following

Hinckley et al. (1996) and Megrey and Hinckley (2001). Egg development was calculated as a

function of age and temperature using 21 development stages (Blood et al., 1994). Eggs hatched

when they reached stage 21 and passed into the yolk-sac larval stage. Yolk-sac larval dry weight

was a function of standard length (Yamashita & Bailey 1989), which was determined by the egg

diameter at hatch (Hinckley 1990). Growth of yolk-sac larvae depended on the number of degree

30

days experienced, which were accumulated every day post-hatch using temperatures from the

physical model for each location and time step. Once the yolk-sac was depleted, if feeding was

successful, surviving larvae entered the feeding larvae stage and a feeding probability was

calculated at each time step. Predation mortality mechanisms are described in section 2.2.2.2.

Dry weight of feeding larvae depended on assimilation efficiency (Houde, 1989), consumption

(modified from MacKenzie et al., 1990), and daily respiration rate (Yamashita and Bailey, 1989).

Feeding rules and the transition from first feeding to feeding larvae were modeled according to

Hinckley, 1999.

Prey consumption by juveniles was estimated utilizing the foraging model for

planktivorous fish implemented by Ciannelli et al. (1998) (Tables 2.1 and 2.2), which was

developed by Bevelhimer and Adams (1993), evaluated by Stockwell and Johnson (1997), and

subsequently field tested by Stockwell and Johnson (1999). For a full description of this

modeling approach, see Eggers (1977) and Stockwell and Johnson (1997, 1999).

A Gerritsen and Strickler (1977) encounter rate model, as simplified by Evans (1989),

was used to estimate the number of each prey type encountered by juvenile pollock per time step

(Eggers, 1977). We assumed that the visual field of foraging pollock was uniform within the

search volume and did not attempt to account for differences in prey position within the visual

field (Mazur and Beauchamp 2006). Reaction distance (cm) was set constant for daytime feeding

and juvenile pollock were only allowed to feed during lighted periods of the diel cycle. In situ

observations of juvenile pollock feeding confirmed that little feeding occurs outside of lighted

periods (Mazur et al. 2007). Prey consumption estimates per time step were generated using a

functional response based on prey encounters rather than weight (cf. Bevelhimer & Adams 1993,

Stockwell & Johnson 1997). The prey field constructed for the model was based on associations

of small copepods, large copepods, and euphausids with bathymetry from field data collected

during 2000 and 2001 cruises (see Table 2.3).

Consumption of each prey type was partitioned based on walleye pollock body length

relative to prey length, and summed to estimate the total number of each prey type potentially

available for consumption per hour (Table 2.3). Prey consumption was not allowed to exceed the

theoretical maximum daily consumption (C-max) for pollock of each length as estimated by the

bioenergetics model (Hanson et al.; 1997; Ciannelli et al., 1998). Consumption of prey during

each time step only occurred when the stomach was not full following digestion.

31

The bioenergetics of juvenile walleye pollock was based on Ciannelli et al.’s, (1998)

model modified for spatially-explicit requirements. Energy consumed (C) was allocated between

metabolism (R), egestion (F), excretion (U), and growth (G) (Hewett and Johnson, 1992):

C=R+F+U+G (1)

Digestion was modeled hourly (Elliott and Pearson 1978) using walleye pollock

evacuation rates from Merati & Brodeur (1996) and Mazur et al. (2007). Numbers of each

potential prey type were converted to weights and added to the ration based on available stomach

capacity and prey preference. Net growth was estimated using the bioenergetics model and the

amount of available prey energy digested on an hourly time step (Stockwell and Johnson 1997).

Net growth estimates (mass, g) were updated each time step and converted to growth in length

(mm) for use in subsequent time steps.

2.2.2.2. Mortality submodel

The daily probability of survival for eggs, yolk-sac larvae, feeding larvae, and juveniles is

characterized as an exponential function dependent on the instantaneous daily mortality rate at the

respective stage. This submodel assumed that each particle represented a batch of eggs released

in a spawning area. Computational constraints restricted the maximum number of particles to

5000 per simulation. Simulations run with a larger number of particles showed the same

retention patterns. To maintain realistic mortality rates within the population, each particle was

considered a superindividual made up of many individuals (Scheffer et al., 1995; Megrey &

Hinckley, 2001) that grew from eggs to juveniles and subject to stage-specific mortality (Table

2.4). Each superindividual contained an initial number of eggs proportional to the egg production

estimated by the walleye pollock stock assessment for that year (Dorn et al., 2005). The number

of surviving superindividuals was assessed at each time step. Starvation mortality was also

included, by removing those individuals whose weight fell below a stage- and size-specific

critical level.

2.2.2.3. Movement submodels

Each batch of individuals that was assigned to a particle which moved according to the u,

v, and w velocity components in a Lagrangian pathway. Particle tracking was based on the Euler

method. Particle trajectories were monitored using a Java tool aligned with the ROMS native

32

grid (Lett et al., 2008) resulting in a unique combination of growth, distribution, and survival for

each superindividual.

Horizontal and vertical movements of individuals were stage specific. The vertical

position of each egg depended on the terminal velocity and the vertical component of water

velocity, w, from the ROMS model at each time step. The terminal velocity was calculated using

Sundby (1983) when Reynolds numbers (Re) were less than 0.5. Yolksac larvae were assumed to

remain at the depth of hatch until first-feeding. Feeding larvae began diel migrations at 6 mm,

with swimming speeds a function of length (Kendall et al., 1987; 1994). Larvae reached their

maximum depth at midday and minimum depth at dusk. Larvae were deeper in the water column

at night compared to crepuscular periods (i.e. dusk or dawn). Horizontal position of eggs and

larvae were mainly influenced by the u and v components of the fluid field.

New algorithms were developed for vertical and horizontal movements of juveniles and

added to the original IBM of Hinckley et al. (1996). Vertical positions of juveniles at each time

step, 1+tZj (m), were calculated based on the depth at the previous time step ( tZj ), the vertical

mean velocity ( jw ), and the time step ( tΔ ).

twZjZj jtt Δ+=+1 (1)

The mean magnitude of jw depended on the length of walleye pollock juveniles (Table

2.2) and was sampled randomly from a triangular distribution1 to select a value at each time step.

The direction of jw (upward or downward) was a random variable. The final depth (m) of each

particle at each time step was bounded according to the hour (h) of the day in a 24 hour cycle

(Equations 2 to 6).

h < 2 40 < 1+tZj < 60 (2)

1 The triangular distribution is a continuous probability distribution with lower limit a, mode c and upper

limit b.

33

2 < h < 5 40 < 1+tZj < 110 (3)

5 < h < 14 90 < 1+tZj < 110 (4)

14 < h < 18 40 < 1+tZj < 110 (5)

h > 18 40 < 1+tZj < 60 (6)

Horizontal positions of juveniles were determined using a correlated random walk based

on Kareiva and Shigesada (1983). We modeled trajectories as a sequence of straight line moves

in which juvenile displacement depended on juvenile length. The position of the juvenile

superindividual at each time step, 1+tXj and 1+tYj , depended on the position in the previous time

step, tXj and tYj the length of the juvenile Lj , and the turning angle , 1+tα . Turning angles were

measured relative to the previous direction of movement using:

jtt θαα +=+1 (7)

where 1+tα is the turning angle at time t + 1, αt was the turning angle at the previous time step,

and the angle jθ was chosen from a normal random probability distribution

)cos( 11 ++ += ttt LjXjXj α (8)

)sin( 11 ++ += ttt LjYjYj α (9)

2.2.3. Model simulations

Two simulation experiments were run to: (1) examine spatial and temporal matching

between model predictions and observed distributions of successive life stages of walleye pollock

in the GOA during 1987, and to (2) predict potential nursery areas in the GOA using model runs

from multiple years.

2.2.3.1. Spatial and temporal matching of model predictions and data

The objective of this experiment was to use the coupled model to hindcast distributions of

larvae in May, late larvae and early juveniles in June/July, and later juveniles in August-

September for the year 1987. This was during a time period when spawning in Shelikof Strait

was the dominant source of eggs and larvae for the GOA. Model output was compared with

results from young larval survey distributions from 18 to 29 May of that year, late larvae and

young juvenile survey distributions sampled between 18 June and 16 July, and late juvenile

34

survey distributions from 12 August to 20 September (Hinckley et al., 1991). Initial conditions

for the simulation (egg locations) were based on the distribution of walleye pollock eggs found

during an April survey (EcoFOCI, Hinckley et al., 1991) in Shelikof Strait. Initial numbers of

eggs corresponded to egg production estimates from the 1987 stock assessment (Table 2.5). Eggs

were released in Shelikof Strait (area 11 in Fig.1.11). The model was run and the distribution of

surviving larvae in May, late larvae and early juveniles in June/July, and later juveniles in

August/September was tabulated. The location and timing of each walleye pollock stage was

compared to those observed during the surveys.

2.2.3.2. Prediction of potential nursery areas

To identify potential nursery areas (PNA) for walleye pollock spawned in the Gulf of

Alaska, the IBM was run for a suite of 6 years: 1978, 1982, 1988, 1992, 1999, 2001 (see Section

2.1). Eggs were released in 14 initial spawning areas (Fig. 1.11), identified as potential spawning

areas, over 4 months in spring: Feb, Mar, Apr, May, which spans the bulk of known walleye

pollock spawning in the GOA (M. Dorn, Alaska Fisheries Science Center, Seattle, WA, pers.

comm.). The model was run for all parameter conditions described above, initializing the number

of eggs according to the egg production estimates from the stock assessment for each year (Table

2.4). At the end of the simulation year, locations of juveniles at day of the year (DOY) 215 were

recorded and PNAs were observed based on areas of surviving juvenile concentration.

2.3. Results

2.3.1. Comparison between observed and predicted distributions

The observed distribution of early walleye pollock larvae in May 1987 showed high

densities in a patch east of the Sutwik and Semidi Islands, 130 km downstream of the spawning

region in Shelikof Strait (Fig. 2.2a). Although the survey area was restricted (Fig. 2.2b), the

model output essentially matches the timing and location of maximum density of early larvae

observed. One notable difference was that the predicted distribution showed a larger area of

moderately high densities of early larvae to the east of the Semidi Islands. Modeled distributions

of late larval and early juvenile walleye pollock in June and July, 1987 were concentrated

between the Semidi and the Shumagin Islands (Fig. 2.2c). The survey data contained high density

peaks in the same area (Fig. 2.2d) during this period, and also showed walleye pollock to the east

of this region. The modeled distribution of later 0-age juvenile walleye pollock in August and

35

September in the GOA was located around the Shumagin Islands (Fig. 2.2e) with highest

densities to the east of the Shumagin Islands. Maximum densities in the survey data were found

to the west of the Shumagins (Fig. 2.2f), not far west of the peak of the modeled densities. The

model predicted no significant numbers of walleye pollock juveniles to the east of this general

area, however small areas of concentration near Sutwik Island and on the northeast side of

Kodiak Island were seen in the data. The survey data (Fig. 2.2f) indicated that juvenile walleye

pollock were found in Unimak Pass, possibly transiting to the Bering Sea. The modeled

distribution at this time (Fig. 2.2e) indicated movement of juveniles to the Bering Sea. In

summary, the model predicted the timing and locations of different life stages of walleye pollock

released from Shelikof Strait that matched those observed in the 1987 survey data, with minor

location differences.

To examine whether modeled walleye pollock juveniles that ended up in the area to the

east of the Shumagin Islands (area 17, Fig. 1.11) could have been spawned in regions other than

Shelikof Strait, we examined the spawning location of all surviving juveniles in area 17 during

1987 (see Figure 1.11 for source-sink regions). The number of fish that arrived in area 17 from

each potential spawning area for all fish spawned in March is shown in Figure 2.3a. The majority

of surviving juveniles were released in area 6, offshore from the Kenai Peninsula between Prince

William Sound and Kodiak Island. Smaller numbers originated in Areas 9 (east of Kodiak

Island), 11 (Shelikof Strait), 12 (southeast of Kodiak Island) and Area 15 (Trinity Island region).

Of those fish released in April (Fig. 2.3b), the major contributors to surviving juveniles were

Areas 6, 9, 11 and 12. The Shelikof Strait spawning contribution to area 17 was predicted to be

more important in March, when spawning actually occurs, than in April. In this model

experiment, fish were released in a stratified random manner over a broad geographic range

within the GOA; we interpret these results as illustrating where fish “could” have been spawned.

2.3.2. General patterns of juvenile pollock nursery areas

The distribution of surviving juveniles modeled for 1987, from simulations using

particles released from 14 spawning areas in the GOA during March, April and May is

summarized in Figure 2.4a. The model domain was subdivided into three main distributional

areas: Gulf of Alaska (GOA), Aleutian Islands (AL), and Bering Sea (BS). A surprising result

(Fig. 2.4b) was the concentration of juvenile walleye pollock in the central and southern Bering

Sea (Areas 33, 34, 36, 37, 38, 41, see Fig. 1.11) that were spawned in the GOA. The overall

36

percentage of GOA spawned fish found on DOY 215 in the BS was 67.7%. The main area for

surviving juveniles in the GOA was the east Shumagin Island region (area 17, Fig. 1.11). Other

areas of concentration in the GOA during 1987 included the area northeast of Kodiak Island, and

near the Semidi Islands (Fig. 2.4b). Overall, 20% of GOA spawned fish remained in the GOA.

The AL region (area 28) contained moderately high densities of surviving juveniles comprising

12.3% of the original number spawned in the GOA.

PNAs varied in location among the six simulation years. The Semidi and Shumagin

Islands areas had consistent high concentrations of juvenile walleye pollock in all years,

especially in 1978, 1982, 1988 and 2001 (Fig. 2.5). Northeast and southeast of Kodiak appear to

be important PNAs in 1978, 1992, and 2001. The BS is a recurring PNA with high densities of

surviving juveniles during all years, although it was less important in 1988 and 1999, when

juvenile densities were lower there compared to the GOA and the AL. Concentrations of juvenile

walleye pollock were predicted to occur in the AL region during 1999 and 2001. Averaging over

all years, the PNA pattern was characterized by a large region in the central and southern BS,

areas east of Kodiak Island and around the Semidi to Shumagin Islands in the GOA, and one area

in the AL (Fig. 2.6).

2.4. Discussion

The location and timing of the maximum densities of early and late larvae, and early

juveniles matched the corresponding distributions seen in the 1987 survey data. The location of

the maximum density of late juveniles predicted by the model did not exactly match the location

of the distribution of late juvenile from the survey data. The surviving late juveniles in the model

and the survey data were both found near the Shumagin Islands, but at a finer spatial resolution,

the location of the maximum density was west of the Shumagin Islands in the survey data,

compared to east of the Shumagins as predicted by the model. However, the eastern limit of late

juvenile distribution, ie. to the east of the Shumagin Islands and in a near-coastal band from Wide

Bay to Unimak Pass were almost identical in both the survey data and model predictions. The

agreement between survey data and model predictions based on corresponding data patterns

(sensu Grimm et al., 2005) demonstrate the ability of this biophysical model suite to predict the

location and timing of early life stages of walleye pollock in the GOA. The movement submodel

used for juvenile pollock, based on a correlated random walk, produced a distribution of early

juveniles that agreed with the survey data. Some improvement of the match between survey data

and model predictions could be made by refinements to the swimming algorithm of late juveniles

37

by including factors such as swimming directionality, however this level of refinement may not

be necessary to obtain a good picture of juvenile distribution.

Two other results from the model predictions were striking: 1) the model correctly predicted

the Shumagin Islands area as the most consistent potential nursery area in the GOA, and 2)

surviving modeled juveniles in the Shumagin Islands could have been spawned in Shelikof Strait

during April (the model experiment included spawning areas which were not necessarily

observed via surveys for each year). This prediction is consistent with literature studies and

confirms the idea of the Shelikof spawning area-Shumagin Island nursery area pair (Hinckley et

al. 1991). Model predictions of nursery areas northeast of Kodiak Island and around the Semidi

Islands are also supported in the literature. A portion of the fish retained around Kodiak Island

(Wilson, 2000), may be advected into the Alaskan Stream and lost from the population (Bailey et

al., 1999), or transported to other nursery areas such as the Semidi Islands (Mazur et al., 2007).

There is no field evidence to support this speculation other than results from parasite studies that

found juvenile walleye pollock found in bays east of Kodiak Island (Fig. 2.1) did not originate in

Shelikof Strait, unless there were anomalous currents (Bailey et al., 1999); implying that areas

other than Shelikof Strait may serve as spawning areas.

The Bering Sea was predicted to be an important potential nursery (or loss) area. This

result has implications for management of walleye pollockin the GOA and the BS as the two are

currently managed as separate stocks. This issue, plus an assessment of whether the GOA

pollock populations should be managed as a single stock, will be further assessed and discussed

in Chapter 3, which describes the full spawning-nursery area connectivity study.

There were potential constraints implied in our combination of the IBM with this

physical flow model. One constraint was the limitations of the hydrodynamic model. The

ROMS output used in these simulations had a 10 km resolution near the coast and did not

accurately characterize physical processes that operated at finer scales. A second potential

constraint was the accuracy of the bathymetry and its influence on water movement. Bottom

topography in the ROMS model grid used (NEP) was smoothed in some regions (e.g. depths in

Shelikof Strait averaged 100 m in the model while actual bottom depths are more like 200 m with

some areas of 300 m). This may influence flow patterns in coastal areas. Recent model-data

comparisons between the ROMS output and GOA oceanographic data indicate that the ROMS

model output includes many circulation features in the northeast Pacific, including variable flow

38

of the Alaska Coastal Current and sea surface height (Hermann et al., In review). Given the

independent validation, we found that the IBM coupled model retained sufficient oceanographic

features to allow corroboration of our model predictions with early life history survey data.

2.5. Literature Cited Bailey, K.M., Bond, N.A., Stabeno, P.J., 1999. Anomalous transport of walleye pollock larvae linked to ocean and atmospheric patterns in May 1996. Fisheries Oceanography 8: 264-273. Bailey, K.M. 2000. Shifting control of recruitment of walleye pollock Theragra chalcogramma after a major climatic and ecosystem change. Mar Ecol Prog Ser 198:215-224. Beck, M.W., Heck, K.L., Able, K.W. Jr., Childers, D.L., Eggleston, D.B., Gillanders, B.M., Halpern, B., Hays, C.G., Hoshino, H., Minello, T.J., Orth, R.J., Sheridan, P.F., and Weinstein, M.P. 2001. The identification, conservation, and management of estuarine and marine nurseries for fish and invertebrates. BioScience 51: 633-641. Bevelhimer, M.S., S.M. Adams. 1993. A bioenergetic analysis of diel vertical migration by kokanee salmon, Oncorhynchus nerka. Can. J. Fish. Aquat. Sci. 50: 2336-2349.

Blood D.M., Matarese, A.C., Yoklavich, M.M., 1994. Embryonic development of walleye pollock Theragra chalcogramma, from Shelikof Strait, Gulf of Alaska. Fish. Bull. US 92, 207-222. Ciannelli, L., Brodeur, R.D., Buckley, T.W., 1998. Development and application of a bioenergetics model for juvenile walleye pollock. J. Fish Biol. 52, 879–898. Curchitser, E.N., Haidvogel, D.B., Hermann, A.J., Dobbins, E.L., Powell, T., Kaplan, A., 2005. Multi-scale modeling of the North Pacific Ocean: Assessment of simulated basin-scale variability (1996-2003). J. Geophys. Res. 110, C11021, doi:101029/2005JC002902. Deegan, L.A., 1993. Nutrient and energy transport between estuaries and coastal marine ecosystems by fish migration. Can. J. Fish. Aquat. Sci. 50, 74-79. De Robertis, A., Ryer, C. H., Veloza, A. and Brodeur, R. D. (2003). Differential effects of turbidity on prey consumption of piscivorous and planktivorous fish. Canadian Journal of Fisheries and Aquatic Sciences 60, 1517–1526. Dorn. M., Aydin, K., Barbeaux, S., Guttormsen, M., Megrey, B., Spalinger, K., Wilkins, M., 2005. Assessment of walleye pollock in the Gulf of Alaska In: Appendix B. Stock assessment and fishery evaluation report for the Groundfish Resources of the Gulf of Alaska, North Pacific Fishery Management Council, P.O. Box 103136, Anchorage, AK 99510. Dumont, H. J., I. Vande Velde, S. Dumont. 1975. The dry weight estimate of biomass in a selection of cladocera, copepoda and rotifera from the plankton, periphyton and benthos of continental waters. Oecologia 19: 75-97.

Eggers, D. M. 1977. Nature of Prey Selection by Planktivorous Fish. Ecology 58(1): 46-59.

39

Elliott, J. M., W. Davison. 1975. Energy equivalents of oxygen consumption in animal energetics. Oecologia 19(3): 195-201.

Elliott, J. M., L. Persson. 1978. The estimation of daily rates of food consumption for fish. J. Anim. Ecol. 47: 977-991.

Evans, G.T. 1989. The encounter speed of moving predator and prey. Journal of Plankton Research. 11(2): 415-417.

Gerritsen, J., J. R. Strickler. 1977. Encounter probabilities and community structure in zooplankton: a mathematical model. J. Fish. Res. Bd. Can. 34: 73-82. Grimm, V., Revilla, E., Berger, U., Jeltsch, F., Mooij, W.M., Railsback, S.F., Thulke, H., Weiner, J., Wiegand, T. and D.L. DeAngelis. 2005. Pattern-Oriented Modeling of Agent-Based Complex Systems: Lessons from Ecology. Science, 310(5750): 987 – 991. DOI: 10.1126/science.1116681 Haidvogel D.B., Wilkin J.L., Young R.E., 1991. A semi-spectral primitive equation ocean circulation model using vertical sigma and orthogonal curvilinear horizontal coordinates. J. Comput. Phys. 94,151-185. Harden Jones, F.R. 1968. Biological aspects of fish migration. Chapter 2 In: Fish Migration, Edward Arnold Ltd. London. Hermann A.J., Stabeno P.J. 1996. An eddy-resolving model of circulation on the western Gulf of Alaska shelf. 1. Model development and sensitivity analyses. J. Geophys. Res. 101(CI), 1129-1149. Hermann, A.J., Curchitser, E.N., Dobbins, E.L., Haidvogel, D.B. 2008. A comparison of remote versus local influence of El Nino on the coastal circulation of the Northeast Pacific. Deep Sea Research II, in review. Hewett, S.W., Johnson, B.L., 1992. Fish Bioenergetics Model 2, an upgrade of ‘ A generalized bioenergetics model of fish growth for microcomputers. ’ University of Wisconsin Sea Grant Technical Report Number WIS-SG-92-250, 79 pp. Hinckley, S., Bailey, K.H., Picquelle, S.J., Shumacher, J.D., Stabeno, P.J., 1991. Transport, distribution, and abundance of larval and juvenile walleye pollock (Theragra chalcogramma) in the w western Gulf of Alska. Can. J. Fish. Aquat. Sci. 48, 91-98. Hinckley S., Hermann A.J., Megrey B.A. 1996. Development of a spatially explicit, individual-based model of marine fish early life history. Mar. Ecol. Prog. Ser. 139, 47-68. Hinckley, S., Hermann, A.J., Mier, K.L., Megrey, B.A., 2001. Importance of spawning location and timing to successful transport to nursery areas: a simulation study of Gulf of Alaska walleye pollock. ICES J. Mar. Sci. 58, 1042-1052. Houde, E.D., 1989. Comparative growth, mortality, and energetic of marine fish larvae: temperature and implied latitudinal effects. Fish. Bull. US 87,471-495.

40

Kareiva, P.M., Shigesada, N., 1983. Analyzing insect movement as a correlated random walk. Oecologia. 56,234-238. Kendall, A.W., CIarke, M.E., Yoklavich, M,M., Boehlert, G.W.,1987. Distribution, feeding, and growth of larval walleye pollock, Theragra chalcogramma, from Shelikof Strait, Gulf of Alaska. Fish. Bull., US 85,499-521. Kendall, A.W., Incze, L.S., Ortner, P.B., Cummings, S.R., Brown, P.K., 1994. The vertical distribution of eggs and larvae of walleye pollock (Theragra chalcogramma) in Shelikof Strait, Gulf of Alaska. Fish. Bull. US 92,540-554. Lett, C., Verley, P., Mullon, C., Parada, C., Brochier, T., Penven, P., Blake, B., (2008). A Lagrangian tool for modelling ichthyoplankton dynamics. Environmental Modelling & Software, 23(9):1210-1214. Link J., T. A. Edsall. 1996. The effect of light on lake herring (Coregonus artedi) reactive volume. Hydrobiologia 332: 131-140. MacKenzie, B.R., Leggett, W.C., Peters, R.H., 1990. Estimating larval fish ingestion rates: can laboratory derived values be reliably extrapolated to the wild? Mar. Ecol. Prog. Ser. 67, 209-225. Mazur, M. M., D. A. Beauchamp. 2006. Linking piscivory to spatial-temporal distributions of pelagic prey fishes with a visual foraging model. Journal of Fish Biology 69: 151-175.

Mazur, M.M., Wilson, M.T., Dougherty, A.B., Buchheister, A, Beauchamp, D.A., 2007. Temperature and prey quality effects on growth of juvenile walleye pollock Theragra chalcogramma: a spatially explicit bioenergetics approach. J. Fish. Biol. 70, 816-136. Megrey, B.A., Hinckley, S. 2001. Effect of turbulence of larval fishes: a sensitivity analysis using an individual-based model. ICES J. Mar. Sci. 58,1015-1029. Merati, N. & Brodeur, R. D. (1996). Feeding habits and daily ration of juvenile walleye pollock, Theragra chalcogramma, in the western Gulf of Alaska. In Ecology of juvenile walleye pollock, Theragra chalcogramma (Brodeur, R. D., Livingston, P. A., Loughlin, T. R. & Hollowed, A. B. eds), pp.65-79. U.S. Department of Commerce, NOAA Technical Report NMFS 126.

Parada, C., Hinckley, S., Horne, J., Dorn, M., Hermann, A., Megrey, B. Comparing simulated walleye pollock recruitment indices to data and stock assessment models from the Gulf of Alaska. Accepted. Mar. Ecol. Prog. Ser Scheffer, M., Baveco, J. M., DeAngelis, D. L., Rose, K. A., Van Nes, E. H. 1995. Super-individuals: a simple solution for modelling large populations on an individual basis. Ecol. Mod. 80, 161–170. Schumacher, J.D., Kendall, A.W. Jr., 1991. Some interactions between young pollock and their environment in the western Gulf of Alaska. Calif. Coop. Oceanic. Fish. Invest. Rep. 32, 22-40. Spring. S., Bailey, K.M., 1991. Distribution and abundance of juvenile pollock from historical shrimp trawl surveys in the western Gulf of Alaska. AFSC Processed Report 91-18, 66p. Alaska Fish. Sci. Cent., Natl. Mar. Fish. Serv., NOAA, 7600 Sand Point Way NE, Seattle WA 98115.

41

Sogard, S. M., Olla, B. L. 2000. Endurance of simulated winter conditions by age-0 walleye pollock: effects of body size, water temperature and energy stores. Journal of Fish Biology 56, 1-21.

Stabeno P.J., Hermann A.J. 1996. An eddy resolving model of circulation on the western Gulf of Alaska shelf. II. Comparison of results to oceanic observations. J. Geophys. Res. 101: 1151-1161. Stockwell, J.D., Johnson, B.M. 1997. Refinement and calibration of a bioenergetics-based foraging model for kokanee. Can. J. Fish. Aquat. Sci. 54: 2659-2676. Stockwell, J.D., and B.M. Johnson. 1999. Field evaluation of a bioenergetics-based foraging model for kokanee (Oncorhynchus nerka). Can. J. Fish. Aquat. Sci. 56(Suppl. 1): 140-151.

Sundby. S., 1983. A one-dimensional model for the vertical distribution of pelagic fish eggs in the mixed layer. Deep Sea Res. 30, 645-661. Svetlichny, L.S. and E.S. Hubareva. 2005. The energetics of Calanus euxinus: locmotion, filtration of food and specific dynamic action. Journal of Plankton Research 27: 671-682. Wilson, M.T., Brodeur, R.D., Hinckley, S., 1996. Distribution and abundance of age-0 walleye pollock, Theragra chalcogramma, in the western Gulf of Alaska during September 1990. In: Brodeur RD, Livingston PA, Loughlin TR, Hollowed AB (eds) Ecology of Juvenile Walleye pollock , Theragra chalcogramma, U.S. Dep. Commer., NOAA Tech. Rep. NMFS 126 p 11-24. Wilson, M.T., 2000. Effects of year and region on the abundance and size of age-0 walleye pollock, Theragra chalcogramma, in the western Gulf of Alaska, 1985-1988. Fish. Bull., U.S. 98, 823-834. Wilson, M.T., Jump, C. and J.T., Duffy-Anderson. 2006. Comparative analysis of the feeding ecology of two pelagic forage fishes: capelin Mallotus villosus and walleye pollock Theragra chalcogramma. Marine Ecology Progress Series, 317: 245-258. Winter, A., Swartzman, G., Ciannelli, L., 2005. Early- to late- summer population growth and prey consumption by age-0 pollock, in two years of contrasting pollock abundance near the Pribilof Islands, Bering Sea. Fisheries Oceanography. 14(4): 307-320. Yamashita, Y., Bailey, K.M., 1989. A laboratory study of the bioenergetics of larval walleye pollock, Theragra chalcogramma. Fish. Bull. US 87,525-536.

42

2.6. Tables

Table 2.1. Variables of the bioenergetic model used for juvenile walleye pollock. (Ciannelli et al. 1998)

Variables of bioenergetic model Functions Consumption

PTfWAC cBc )(⋅=

)1(()( VXX eVTf −= )/()( cocmcm TTTTV −−=

400/))/401(1(( 25.02 YZX ++= ))(ln( cocmc TTQZ −= )2)(ln( +−= cocmc TTQY

Respiration

)())(( FCDAmTfWAR sBr

r −+=

)1(()( VXX eVTf −= )/()( rormrm TTTTV −−=

400/))/401(1(( 25.02 YZX ++= ))(ln( rormr TTQZ −= )2)(ln( +−= rormr TTQY

convjRR j ⋅= Egestion CFF a= Excretion )( FCUU a −=

43

Table 2.2. Parameters for bioenergetic and feeding model of juvenile walleye pollock.

Parameter description and unit Symbol Value Reference Consumption (gg-1 day-1) C Proportion of maximum consumption P 0-2 1 Intercept of the allometric function Ac 0.38 1 Slope of the allometric function Bc 0.68 1 Temperature dependence coefficient Qc 2.6 1 Optimum temperature for consumption (oC) Tco 10 1 Maximum temperature for consumption (oC) Tcm 15 1 Respiration (gg-1 02 day-1) R Intercept of the allometric function Ar 0.0075 1 Slope of the allometric function Br 0.251 1 Temperature dependence coefficient Qr 2.6 1 Optimum temperature for respiration (oC) Tro 13 1 Maximum temperature for respiration (oC) Trm 18 1 Proportion of assimilated energy lost for Specific Dynamic Action Ds 0.125 1 Multiplier for active metabolism Am 1 1 Respiration in Joules Rj Conversion from g of oxygen to joules convj 13560 2 Egestion F Proportion consumed energy Fa 0.15 1 Excretion U Proportion of assimilated energy Ua 0.11 1 Digestion (gg-1day-1) D Stomach capacity 3 Digestion coefficient dc 0.25 1 Parameters of feeding model Pollock swimming speed (m s-1) Pss 0.15 4 Euphausid swimming speed (m s-1) Ess 0.0285 16 Large copepod swimming speed (m s-1) Lcss 0.01 17 Small copepod swimming speed (m s-1) Scss 0.002 17 Reactive distance (m) rd 0.1 5 Handling time (seconds prey-1) th 0.33 6 Minutes to hours conv 60 - Probability of prey preference Euphausids PpE 0 (size class 1)** 14 0.1429 (size class 2)** 14

44

0.8571 (size class 3)** 14 Large copepods PpLc 0.127 (size class 1)** 14 0.2381 (size class 2)** 14 0.6349 (size class 3)** 14 Small copepods PpSc 0.6481 (size class 1)** 14

0.32419 (size class 2)** 14

0.02781 (size class 3)** 14 Mean weight Euphausids (g) MwE 0.0402 7 Mean weight Large Copepods(g) MwLc 0.00056743 8 Mean weight Small Copepods(g) MwSc 0.0001122 8 Energy euphausid EnE 5949 3 Energy large copepod EnLc 5319 3 Energy small copepod EnSc 3040 3 Transferemce energy grams pollock 4050 3

** Size class correspond to 1: 25 <= length < 40, 2: 120 > length >= 40 and 3: length >= 120 juvenile walleye pollock.

Variables of Feeding model Functions References

Total prey available per juvenile pollock PpScnumScPpLcnumLcPpEnumETp ⋅+⋅+⋅= 9 Number euphausids consumed per hour convtSvEPpEnumESvEnEch h ⋅⋅+⋅⋅= )1/(( 10, 6 Number large copepods consumed per hour convtSvLcPpLcnumLcSvLcnLcch h ⋅⋅+⋅⋅= )1/(( 10, 6 Number small copepods consumed per hour convtSvScPpScnumScSvScnScch h ⋅⋅+⋅⋅= )1/(( 10, 6 Weight euphausids consumed per hour MwEnEchwEch ⋅= Weight large copepods consumed per hour MwLcnLcchwLcch ⋅= Weight small copepods consumed per hour MwScnScchwScch ⋅= Search volume for euphausids )( 222 EssPsssqrtrdPISvE +⋅⋅⋅= 11 Search volume for large copepods )( 222 LcssPsssqrtrdPISvLc +⋅⋅⋅= 11 Search volume for small copepods )( 222 ScssPsssqrtrdPISvSc +⋅⋅⋅= 11 Stomach capacity per hour DhTpScSc −+= )( 1

Digestion per hour )

)1()((()( )dc

dc

edcTconseScTconsScDh −

−

−⋅+⋅−+=

12, 13 W=weight (g), T=temperature (oC) (1) Ciannelli et al. (1998), (2) Elliott and Davidson (1975), (3) Mazur et al. (2007), (4) Sogard & Olla (2002), (5) Link & Edsall (1996), (6) Stockwell & Johnson (1997), Winter el al. (2005), (8) Dumont et al. (1979), (9) Eggers (1977), (10) Gerritsen and Strickler (1977), (11) Evans (1989), (12) Elliott & Persson (1978), (13) Bevelhimer & Adams (1993), (14) Wilson et al. (2006), (15) Hinckley, 1999, (16) De Robertis et al. (2003), (17) Svetlichny and Hubareva (2005).

45

Table 2.3. Values and functions of prey field (num m-3) assigned for the feeding model associated to different environmental conditions for juvenile walleye pollock.

Prey category Variable Environmental condition Value (number m-3) Reference

Euphausids numE h < 1000 m 1.178 Wilson et al., (2006) h >= 1000 m None jd < 120 ** 0.2946 Hinckley, 1999

160 >=jd >=120**

3566.2)24

(0221.0 −+=itjdnumE

Hinckley, 1999 jd > 160** 0.2946 Hinckley, 1999 Large copepods numLc h < 1000 m 66.663 Wilson et al., (2006) h >= 1000 m None jd < 120** 16.666 Hinckley, 1999

160 >=jd >=120**

55.972)24

(1177.9 −+=itjdnumLc

Hinckley, 1999 jd > 160** 16.666 Hinckley, 1999 small copepods numSc h < 1000 m 486.277 Wilson et al., (2006) h >= 1000 m None jd < 120 ** 121.569 Hinckley, 1999

160 >=jd >=120**

33.133)24

(2499.1 −+=itjdnumSc

Hinckley, 1999 jd > 160** 121.569 Hinckley, 1999

h = bathymetry, jd = julian day, **In the Semidi region see Fig. 2.1.

46

Table 2.4. Mortality rates for egg, larvae and juveniles of walleye pollock used for IBM

experiments.

Daily mortality rates

Year Eggs* Feeding larvae* Juveniles** 1987 0.226 0.064 0.00005 1988 0.300 0.036 - 1989 0.170 0.157 - 1990 0.150 0.073 0.01400 1991 0.220 0.126 - 1992 0.184 0.049 - 1993 - 0.038 0.00607 1994 - 0.057 - 1996 - 0.037 0.01076 1999 - - 0.00157

2001 - - 0.00353 Other 0.205*** 0.0200*** 0.00509

*For egg and larvae daily mortality rates for pollock and methods of calculations, see Bailey et al. 2000, ** Juveniles daily mortality was inferred from stomach contents from groundfish consumption and normalized to a maximum mortality, for missing values we used an average of the available data, *** average values for missing years (Bailey pers. comm.),

Table 2.5. Egg production for stock assessment estimates Dorn et al., (2005) for years of the IBM simulation. Year Egg production 1978 1.593*1014 1988 1.194*1014 1999 5.513*1013 1982 1.677*1014 1992 8.62*1013 2001 5.01*1013

47

2.7. Figures

Figure 2.1. The western Gulf of Alaska.

48

Figure 2.2. Distribution of early walleye pollock larvae in May 1987 a) model and b) data. Distribution of late larval and early juvenile walleye pollock in June and July 1987 c) model and d) data. Distribution of late juvenile walleye pollock in August and September 1987 e) model and f) data. Areas with no larvae are white.

a b

c d

e f

49

Figure 2.3. Density of juvenile walleye pollock that were found on DOY 215 in area 17 (the east Shumagins) 1987 by spawning region and date of release a) March and b) April.

0

500000

1000000

1500000

2000000

2500000

3000000

3 5 6 8 9 11 12 14 17

Origen Area

Surv

ival

juve

nile

s

Where succesful individuals at Shumagin NA comes from release in April?

Release March vs April : successful one at West Shumagin Nursery area

0

500000

1000000

1500000

2000000

2500000

3000000

3 5 6 8 9 11 12 14 15 17

Origen Area

Surv

ival

juve

nile

sWhere succesful individuals at Shumagin NA comes from release in March?

a

b

50

Figure 2.4. a) Contour map of modeled juvenile density at DOY 215 in 1987 showing potential nursery areas through the whole domain. b) Modeled juvenile density at DOY 215 in 1987 discretized by region: Gulf of Alaska (GOA), Aleutians and Bering Sea (BS).

020000000400000006000000080000000

100000000120000000140000000

2 3 5 6 7 8 9 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 28 29 30 32 33 34 35 36 37 38 39 40 41 42 44 45 46

Destiny Areas

Juve

nile

sur

viva

l

GOA BSAleutians

West Shumagin

Outer Southern BS

Central Southern BSb

a

51

Figure 2.5. a) Contour map of modeled juvenile density on DOY 215 showing potential nursery areas through the whole domain in years a) 1978, b) 1982, c) 1988, d) 1992, e) 1999, and f) 2001.

a b

c d

e f

52

Figure 2.6. Contour map of modeled juvenile density on DOY 215 showing potential nursery areas throughout the whole domain averaged over all years of the simulation.

53

CHAPTER 3: Connectivity of Walleye Pollock (Theragra chalcogramma) Spawning and Nursery Areas in the Gulf of Alaska

(Draft Manuscript – Citation is not allowed without author permission) C. Paradaa,c, S. Hinckleyb, J. Hornec aJoint Institute for the Study of Atmosphere and Ocean (JISAO) University of Washington, Box 355672, Seattle, WA 98105 bAlaska Fisheries Science Center, National Oceanographic and Atmospheric Administration (NOAA), 7600 Sand Point Way NE, Seattle, WA, 98115 cSchool of Aquatic and Fishery Sciences, University of Washington, Box 355020, Seattle, WA, 98195 3.1. Introduction

The continuing debate over the efficacy of spatial fishery management necessitates the

development of techniques to explain spatial dynamics of marine populations. The logistical

constraints of tracking small organisms over hundreds of kilometers emphasize the need when

investigating larval dispersal (Willis et al. 2003; Sale et al. 2005). Computer simulation models

have contributed to understanding the kinematics of older life stages, but the dispersal, survival,

and connectivity between spawning and nursery areas remains unresolved for many marine

populations (Beck et al. 2001; Cowen et al. 2007). Since connectivity plays an important role in

local and metapopulation dynamics, community structure, and genetic diversity (Hastings and

Harrison, 1994), understanding the connectivity among life stages within and across populations

provides information to evaluate and design management strategies. The term connectivity is

derived from metapopulation analyses: dynamic interactions between geographically separated

populations via the movement of individuals (North et al., In press). In this study we refer to

connectivity as the dynamic interaction between geographically separated spawning and nursery

areas via the combined effect of individual movement and currents on transport. Connectivity is

defined in practice here as the proportion of age-0 juveniles that are found on September 1st in a

specific nursery area. Nursery areas are defined as the areas of accumulation of age-0 juvenile

that contain more than 5% of survivals from the initial release.

Linkages between populations occur through movements of individuals, either by adult

locomotion or by dispersal of pelagic eggs, larvae, and, to some extent, juveniles (Levins 1969;

Botsford et al. 2001). Almost all fish produce larvae that can spend days, weeks, or months

drifting, eating, and growing in the plankton (Gaines et al. 2007). Corresponding scales of

dispersal from release sites can vary by more than six orders of magnitude, ranging from meters

54

to hundreds of kilometers (Cowen et al. 2000; Gaines et al. 2007; Pineda et al. 2007). In regions

with complex horizontal circulation patterns, trajectories of individual fish may differ widely,

resulting in differing histories of exposure to environmental variables such as temperature,

salinity, and predators and prey, leading to variability in growth and survival among individuals.

Single species populations are often divided into subpopulations, with important connections

among groups, and intermixing of these subpopulations can be related to spatial dynamics of

spawning and nursery areas based on individual movements and currents. Understanding transfers

between spawning and nursery areas provides insight into stock and population structure.

Walleye pollock (Theragra chalcogramma) is a dominant component of the Gulf of

Alaska (GOA) ecosystem, but knowledge of the mechanisms underlying variability in recruitment

is incomplete. Adult walleye pollock are known to spawn from late March to early April at the

southwestern end of Shelikof Strait, between Kodiak Island and mainland Alaska (Kendall et al.,

1987; Schumacher & Kendall 1991). Eggs are fertilized at depths between 150 and 200 m, and

hatch into larvae after a period of about 2 weeks. These larvae rise to the upper 50 m of the water

column and drift in prevailing currents for the next several weeks (late April through mid-May).

Larger larvae undergo diel migrations between 15 and 50 m. Currents may carry the larvae

southwestward along the Alaska Peninsula, or offshore along the shoreward edge of the Shelikof

sea valley southwest of Shelikof Strait. Only a small portion of the larvae hatched from the

spawned eggs survive. By mid-summer, many of the survivors have been advected to the

Shumagin Islands about 300 km southwest of the spawning site (Hinckley et al., 1991). The

prevailing hypothesis is that Shelikof Strait is the main spawning area and the Shumagin Islands

provides the main nursery area in the GOA. Other spawning areas have been observed from

spawner biomass acoustic surveys, but potential contributions to walleye pollock populations in

the GOA by these spawning areas are not known.

Gulf of Alaska walleye pollock are currently managed as a single stock, independent of

Bering Sea and Aleutian Islands walleye pollock. The separation of stocks into eastern Bering

Sea and Gulf of Alaska is based on studies of larval drift from spawning locations (Hinckley et

al., 2001), [s1]and genetic studies of allozyme frequencies (Grant and Utter 1980; Olsen et al.

2002, mtDNA variability (Mulligan et al. 1992; Shields and Gust 1995; Kim et al. 2000), and

microsatellite (O’Reilly et al. 2004) allele variability. It is important to note that data used for the

larval transport study did not include encompass the entire GOA and that genetic analyses have

not provided definitive results on the separation or mixing of genetic population components.

55

The alternate approach of examining connectivity between adult spawning and juvenile retention

areas may contribute to the understanding of walleye pollock population structure. Understanding

linkages between specific spawning locations and the destinations of surviving juveniles may be

used to justify management units of northeast Pacific walleye pollock and aid in the conservation

of population genetic components.

In this study we use a spatially-explicit IBM coupled to a hydrodynamic model to reveal

patterns of potential connectivity of walleye pollock between spawning sites within the GOA, and

retention of surviving juveniles in the GOA, Bering Sea (BS), or the Aleutians (AL). Spatially

explicit IBMs are efficient Lagrangian tracking tools in connectivity studies (Werner et al., 2001,

North et al., In press) and can be paired with geographic information systems (GIS). GIS is used

to delineate source populations and potential nursery habitat along an individual's trajectory. Our

goal is to understand the structure of the walleye pollock populations in Alaskan continental shelf

and slope waters, through modeling the life history from eggs to age-0 juveniles.

3.2. Methods

The modeling approach coupled a biophysical model, the Regional Ocean Modeling

System (ROMS), with a modified individual based model (IBM) of walleye pollock life history

(Hinckley et al, 1996; Megrey and Hinckley, 2001).

3.2.1. Model configuration

A multidecadal (1978-2003) simulation of water currents and temperatures was

conducted for the Gulf of Alaska using the ROMS hydrodynamic model with the 10 km

Northeast Pacific (NEP) grid (Fig. 1.2). Details of this simulation can be found in Curchitser et

al. (2005). The model is a free-surface, hydrostatic primitive equation ocean circulation model

that uses a nonlinear stretched terrain following vertical coordinates. The grid is curvilinear and

pie-shaped from 20.6º to 71.6ºN and from 145.8º to 247.6ºE. Horizontal space is discretized

using orthogonal curvilinear coordinates on an Arakawa C grid (numerical details can be found in

Haidvogel et al. 2000; Moore et al. 2004; Shchepetkin and McWilliams, 2005). The use of 30

vertical levels ensures high resolution near the surface. At the horizontal boundaries facing the

open ocean, an implicit, active, radiative boundary scheme (Marchesiello et al. 2001), is forced

by seasonal time-averaged outputs from a basin scale ocean model. The model was forced with

56

monthly averaged fluxes of wind, heat, and salinity from the Comprehensive Ocean-Atmosphere

Data Set (COADS) ocean surface climatology, with a horizontal spatial resolution of 0.5º (Da

Silva et al. 1994). The circulation model simulation was started from rest, and summer values

were used for initial conditions. As the model domain is relatively small, the model reached

equilibrium after a spin-up period of about 2 years. The circulation model was run for 50 years

using interannual variability in the forcing fields.

The 6 coupled model run years included 3 years before (1978, 1982, 1988) and 3 years

after (1992, 1999, 2001) the shift in control of recruitment dynamics of walleye pollock in the

GOA characterized by an increase of juvenile pollock mortality due to a gradual build-up of

groundfish predators during the mid- and late-1980s (Bailey et al. 2000). The particular forcing

scenario imposed on the physical model each year consisted of winds, freshwater input, and

boundary conditions for each different year (see Curchitser et al., 2005 for details of

configuration). The hydrodynamic model produced daily averaged output consisting of salinity

and temperature fields, and 3D velocities. These physical variables were used to drive the IBM

model over the same years. The IBM was run independently of the ROMS model runs. Details

of the physical model simulations and the assessment of the model’s ability to reproduce

observed variability and their impact in the northeast Pacific can be found in Curchitser et al.

(2005).

3.2.2. Individual-based model, parameters, and mechanisms

Individual eggs, larvae, and juveniles were treated as particles with vertical (and in the

case of juveniles, horizontal) locomotory capability. Each particle was tracked through space

over time using a Java-based float tracking application that used modeled velocities from ROMS

output at particle locations to compute movements (see Chapter 1 for details). The IBM was run

from February 1st to September 1st in each model year. The biological mechanisms and parameter

values used for this experiment matched those used in Chapter 2, and included early life stage

processes for eggs, yolk-sac and feeding larvae, and age-0 juvenile pollock.

3.2.3 Spatial and temporal variability of spawning areas

To examine potential spatiotemporal variability in the number of surviving juveniles

from spawning locations, we varied the spatial and temporal spawning parameters in the IBM.

57

Table 3.1 shows areas and months of release for each year of the simulation. The timing of egg

release was set to 1st February, 1st March, 1st April, and 1st May. Also shown in Table 3.1 are the

fourteen initial release areas with codes corresponding to the spawning and nursery areas in

Figure 3.1. Names and numbers of all of the areas in the model domain, and the sector associated

with each area are shown in Table 3.2 and Figure 3.2.

The 14 spawning areas selected for the simulations were either areas where spawning has

been observed during surveys of walleye pollock egg distribution, or during acoustic surveys of

GOA walleye pollock biomass (Table 3.3). Table 3.3 lists known or suspected walleye pollock

spawning areas in the Gulf of Alaska, the corresponding name and number of the area in the

model domain, and the most likely spawning time for each area. In Morzhovoi Bay trawl catches

were predominately males, so it is unclear whether this location is actually used as a spawning

area.

3.2.4. Spatial and temporal variability of nursery areas

The term nursery area was defined in this study as regions where age-0 juvenile pollock (

>= 25 mm SL) were concentrated on September 1st and represented more than 5% of all

survivors. The hydrodynamic model domain was divided into 46 smaller areas according to

bathymetry and topography (islands, sounds, passes, bays, straits). Potential nursery areas

included regions where juveniles have been sampled in FOCI surveys or during NMFS acoustic

surveys of the walleye pollock biomass (Table 3.4). The simulated number of surviving 0-age

juveniles in each area was tabulated.

3.2.5. Transport to and retention in nursery areas

To examine contributions of surviving juveniles by spawning areas within the model

domain, the proportion of juveniles from each spawning area was tabulated by month and year.

To examine the relative contributions we calculated two variables: (1) retention and (2) transport

of modeled age-0 juveniles to nursery areas. Retention was defined as the proportion of eggs

released in a spawning area that on the 1st of September was still in the same spawning area.

Transport was defined as the proportion of eggs released in a spawning area that went to area X.

Variable values were initially averaged over all release years and months, and then separated by

month and by year. The objective of this analysis was to identify regions where surviving

juveniles were concentrated and report retention and transport proportions that exceeded 5% of

58

the surviving population. To examine the connectivity between spawning and nursery areas we

created a connectivity matrix by month. We discuss model results relative to what is known or

suspected about walleye pollock spawning and nursery areas.

Table 3.1. Spatial (spawning areas) and temporal (month) release parameters used in the

simulation.

Parameter Number Description

Spawning Area 2 Inner Cook (InC)

3 Prince Williams Sound Inner (PWSin)

5 Outer Cook Inlet(OC)

6 Seward Inner (Sin)

8 Shelikof Strait North (SSN)

9 Kodiak Island North (KIN)

11 Shelikof Strait Exit (SSE)

12 Kodiak Island South (KIS)

14 Semidi Islands (SemI)

15 Sutwik Island (Sut)

17 Shumagin Islands Inner (SIin)

18 Shumagin Islands Outer (SIo)

20 Unimak Pass (UP)

21 Unimak Pass Outer (UPo)

Date of release 1st February

1st March

1st April

1st May

Yeasr of simulation 1978, 1982, 1992, 1999, 2001

59

Figure 3.1. Map of the numbers of the regions used to set initial conditions of release of particles (spawning areas) in South East Alaska (SEA), the Gulf of Alaska (GOA), the Aleutians (AL) and the Bering Sea (BS) sectors. The area numbers and corresponding names are listed in Table 3.2. The spawning regions are Inner Cook (InC), Prince Williams Sound Inner (PWSin), Outer Cook (OC),Seward Inner (Sin), Shelikof Strait North (SSN), Kodiak Island North (KIN), Shelikof Strait Exit (SSE), Kodiak Island South (KIS), Semidi Islands (SemI), Sutwik (Sut), Shumagin Islands Inner (SIin), Shumagin Islands Outer (SIo), Unimak Pass (UP), Unimak Pass Outer (UPo). The name and the corresponding number of the areas where the destiny of particles is counted are listed in Table 2. The areas that where assessed as nursery areas where the same spawning areas plus the areas: South East Alaska (SEA), 1: Yakutat (Yak), Prince Williams Sound Outer (PWSo), Seaward Offshore (So), Kodiak Island North Offshore (KINof), Kodiak Island South Offshore (KISof), Sutwik Offshore (Suto), Shumagin Islands Offshore (SIof), Unimak Pass Offshore (UPof), Unalaska Island (UI), Unalaska Island (UIof), Chagulak Island (CI), Adak (Ad), Cobra Dane (CD), Offshore (Off), Bering Sea South Inner domain (BSSin), Bering Sea South Middle

domain (BSSm), Bering Sea South Outer domain (BSSo), Bering Sea South Basin (BSSb), Bering Sea Central Inner domain (BSCin), Bering Sea Central Middle domain (BSCm), Bering Sea Central Outer domain (BSCo), Bering Sea Central Basin (BSCb), Bering Sea North Inner domain (BSNin), Bering Sea North Middle domain (BSNm), Bering Sea North Outer domain (BSNo), Bering Sea North Basin (BSNb), Arctic Inner domain (Arin), Arctic Middle domain (Arm), Arctic Outer domain (Aro), Arctic Basin (Arb).

60

Figure 3.1b. Map of the numbers of the regions used to set initial conditions of release of particles (spawning areas) in South East Alaska (SEA), the Gulf of Alaska (GOA), the Aleutians (AL), the Bering Sea (BS), and the Arctic region (Ar) sectors. The area numbers and corresponding names are listed in Table 3.2.

Table 3.2. Names and numbers of all areas used in the simulation. The fourteen initial areas of release (spawning areas) are indicated by asterisks.

Region number Region name Sector

0 South East Alaska (SEA) SEA

1 Yakutat (Yak) GOA

2* Inner Cook Inlet (InC) GOA

3* Prince Williams Sound, Inner (PWSin) GOA

4 Prince Williams Sound, Outer (PWSo) GOA

5* Outer Cook Inlet (OC) GOA

61

6* Seward, Inner (Sin) GOA

7 Seward, Offshore (So) GOA

8* Shelikof Strait North (SSN) GOA

9* Kodiak Island North (KIN) GOA

10 Kodiak Island North Offshore (KINof) GOA

11* Shelikof Strait Exit (SSE) GOA

12* Kodiak Island South (KIS) GOA

13 Kodiak Island South Offshore (KISof) GOA

14* Semidi Islands (SemI) GOA

15* Sutwik Island (Sut) GOA

16 Sutwik Island,Offshore (Suto) GOA

17* Shumagin Islands, Inner (SIin) GOA

18* Shumagin Islands, Outer (SIo) GOA

19 Shumagin Islands, Offshore (SIof) GOA

20* Unimak Pass (UP) GOA

21* Unimak Pass, Outer (UPo) GOA

22 Unimak Pass, Offshore (UPof) GOA

23 Unalaska Island (UI) AL

24 Unalaska Island (UIof) AL

25 Chagulak Island (CI) AL

28 Adak (Ad) AL

29 Cobra Dane (CD) AL

30 Offshore (Off) GOA, AL

31 Bering Sea South, Inner domain (BSSin) BS

32 Bering Sea South, Middle domain (BSSm) BS

33 Bering Sea South, Outer domain (BSSo) BS

34 Bering Sea South, Basin (BSSb) BS

35 Bering Sea Central, Inner domain (BSCin) BS

36 Bering Sea Central, Middle domain (BSCm) BS

37 Bering Sea Central, Outer domain (BSCo) BS

38 Bering Sea Central, Basin (BSCb) BS

39 Bering Sea North, Inner domain (BSNin) BS

40 Bering Sea North, Middle domain (BSNm) BS

41 Bering Sea North, Outer domain (BSNo) BS

62

42 Bering Sea North, Basin (BSNb) BS

43 Arctic, Inner domain (Arin) Ar

44 Arctic, Middle domain (Arm) Ar

45 Arctic, Outer domain (Aro) Ar

46 Arctic Basin (Arb) Ar

Gulf of Alaska (GOA), Aleutians (AL), Bering Sea (BS), Arctic (Ar)

Table 3.3. Known or suspected walleye pollock spawning areas in the Gulf of Alaska. The spawning areas and spawn timing in each area, and the corresponding name and number of the spawning areas used in the model experiment are indicated.

Modeled areas corresponding to the

observed spawning areas

Observed Spawning

areas in Gulf of Alaska

Spawning Timing (likely

occurrence)

Region

number Region name

Morzhovoi Bay* Mid to late February 20 Unimak Pass (UP)

Sanak Trough Early to mid February 21 Unimak Pass Outer (UPo)

Shumagin Gully Mid to late February 17 Shumagin Islands Inner (SIin)

Chirikov shelf break Late March to early April 15 Sutwik Island (Sut)

8 Shelikof Strait North (SSN) Shelikof Strait Late March to early April

11 Shelikof Strait Exit (SSE)

Marmot Bay Late March to early April 9 Kodiak Island North (KIN)

Middleton Island Late March to early April 6 Seward Inner (Sin)

Entrance to Prince William Sound

Late March to early April 3

Prince Williams Sound Inner

(PWSin)

* Trawl catches nearly all males, so it is unclear whether this is actually a spawning area.

63

Figure 3.2. Walleye pollock observed spawning areas in the Gulf of Alaska

64

Table 3.4. Known or suspected walleye pollock nursery areas in the Gulf of Alaska. Observed nursery areas and the corresponding region name and number in the model are indicated.

Modeled areas corresponding to the nursery areas

Observed Nursery areas in Gulf of

Alaska *

Region

number Region name

Shumagin area from Semidis islands to

Unimak Pass 14 Semidi Islands (SemI)

15 Sutwik Island (Sut)

17 Shumagin Islands Inner (SIin)

18 Shumagin Islands Outer (SIo)

20 Unimak Pass (UP)

21 Unimak Pass Outer (UPo)

North of Kodiak Island 5 Outer Cook Inlet (OC)

6 Seward Inner (Sin)

North East of Kodiak Island 9 Kodiak Island North (KIN)

Southwest of Unimak Pass 20 Unimak Pass (UP)

* Matt Wilson pers. comm.

65

3.3. Results

3.3.1. Retention in the spawning areas and temporal variability

The maximum mean retention of age-0 juvenile walleye pollock over all years, months,

and all spawning areas by September 1st was 0.28 in the Shumagin Islands Inner area (SIin) .

This was followed by Prince William Sound Inner area (PWSin) and Outer Cook Inlet (OC) area

with retention proportions of 0.22 (Fig. 3.3a). The Seward Inner area (Sin) retained 7% of the

spawned eggs. Retention values did not exceed 5% in other spawning areas (Fig. 3a). For

retention levels > 5%, monthly variability in the release time of spawning showed that the later

the spawning occurred, the higher the proportion of retention. Eggs spawned in February resulted

in the lowest proportion of retention, and those spawned in May showed the highest retention

(Fig. 3b). These results indicate that the highest cumulative mortality was experienced by

individuals spawned in the early months, as indicated by the monotonic increase of numbers over

time, as expected. A contrasting example occurred within the SIin spawning area, however. The

retention of May-spawned age-0 juveniles in September decreased relative to those spawned in

April (Fig. 3b). In the InC spawning area, retention was only observed among eggs released in

April and May, with May showing the highest level of retention (8%) from that site. Across sites,

interannual variability in retention was low, with the highest variability observed within spawning

sites (Fig.3c). Spawning areas SIin, PWSin, OC, and Sin consistently contained the overall

highest retention rates, but locations differed among years (Fig.3c).

3.3.2. Temporal variability of transport to nursery areas

The highest proportion of age-0 juveniles that were alive on September 1st across years

occurred in the Shumagin Islands Inner area (SIin), followed by Semidi Islands (SemI), and Outer

Cook Inlet (OC) (Fig. 3.4). In the Bering Sea, the highest proportion of age-0 juveniles were

found in the Bering Sea South Basin (BSSb) area, followed by the Bering Sea South Outer

domain (BSSo), the Bering Sea Central Outer domain (BSCo), and the Bering Sea Central Middle

domain (BSCm) areas. Timing of egg release affected survival, with transport to SemI, OC, and

BSSo areas increasing from release dates February to May, with simultaneous decreases in BSCm

and BSCo areas. Transport into SIin and BSSb nursery areas increased among eggs released up to

April and then decreased during May (Fig. 3.4). Transport and retention values have not been

corrected for cumulative mortality.

66

0.000.050.100.150.200.250.300.350.40

InC

PW

Sin OC

Sin

SS

N

KIN

SS

E

KIS

Sem

I

Sut

Siin Sio UP

Upo

Spawning areas

Rete

ntio

n

a

0.000.020.040.060.080.100.120.140.16

InC

PW

Sin OC

Sin

SS

N

KIN

SS

E

KIS

Sem

I

Sut

Siin Sio UP

Upo

Spawning areas

Ret

entio

n

FebMarAprMay

b

InC

PW

Sin OC

Sin

SS

N

KIN

SS

E

KIS

Sem

I

Sut

Siin Sio UP

Upo

1978

1982

1988

1992

1999

2001

Spawning areas

Years

0.00-0.10 0.10-0.20 0.20-0.30 0.30-0.40

c

Figure 3.3. Retention of juvenile age-0 walleye pollock in each spawning area. a. Proportion of juvenile age-0 retained in spawning areas where released, over all simulations. b. Proportion of juvenile age-0 retained in spawning areas where released, by month of release (1st February, 1st March, 1st April, 1st May). c. Proportion of juvenile age-0 retained in spawning areas where released, by year of release (1978, 1982, 1988, 1992, 1999, 2001) over all spawning monthsXXXX.

67

Figure 3.4. Proportion of age-0 juveniles that were alive at the end of the simulation (September 1st) in a given nursery area. The bars represent the month of release of the eggs, to observe the temporal variability of nursery areas due to differential timing of the spawning process.

3.3.3. Spawning in February

For eggs released on February 1st, the largest transport to nursery areas was to the SIin in

the GOA and to BSSb in the BS with proportions of 0.12 and 0.11. A group consisting of BSSo,

BSCm, and BSCo in the Bering Sea all had transport proportion values of 0.06 (Fig. 3.5).

To identify where individuals that were transported to nursery areas originated, and

whether there was any monthly variability due to the timing of spawning, we built spawning and

nursery area connectivity matrices over all years and then by month. The most likely nursery area

was the Shumagin Islands inner area (SIin) with particles originating from spawning areas on the

inner side of the continental shelf, upstream of SIin, but with considerable retention also

occurring within SIin. Weaker transport to this area was seen from spawning areas located on the

outer edge of the continental shelf of GOA, including from Seward inner (Sin), Kodiak Island

North (KIN), Kodiak Islands South (KIS), and Sutwik Island (Sut) areas. Retention values of ~

1% were seen in SIin (Fig. 3.6).

The second important February nursery area was located in the Bering Sea South basin

(BSSb) with 11% of transport to that nursery region. Spawning areas where those age-0 juveniles

originated were in the GOA at the outer edge of the continental shelf (Sin, KIN, KIS, Sut, SIo, and

UPo), with transports values ranging between 10 and 25% (Fig.3.6). Age-0 juveniles located in

0

0.05

0.1

0.15

0.2

0.25

InC

PW

Si n OC

Si n So

SSN KIN

KI N

ofS

S E KI S

KIS

ofS

emI

Sut

Sut

oS

i in Sio

Sio

fU

PU

Po

UP

of UI

Uio

f

CI

Ad

CD Of f

BSS

inB

S So

BS S

bB

SCin

BSC

mB

SCo

BS C

bB

S Cin

BSN

mB

SNo

BS N

b

Ari n

Arm Aro

Arb

Nursery areas

Prop

ortio

n

FebMarchAprilMay

68

BBSb nursery area were also spawned along the inner edge of the GOA continental shelf, from

the Shelikof Strait Exit (SSE), the Semidi Islands (SemI), the Shumagin Islands inner (SIin), and

Unimak Pass (UP) areas, with transport proportions ranging between 5 and 10% (Fig. 3.6).

0

0.05

0.1

0.15

0.2

0.25

InC

PWSi

nO

C Sin

SoSS

NKI

NKI

Nof

SSE

KIS

KISo

fSe

mI

Sut

Suto

Siin

Sio

Siof UP

UPo

UPo

fU

IU

iof

CI

Ad CD

Off

BSSi

nBS

SoBS

SbBS

Cin

BSC

mBS

Co

BSC

bBS

Cin

BSN

mBS

No

BSN

bAr

inAr

mAr

oAr

b

Prop

ortio

n Feb

a

0

0.05

0.1

0.15

0.2

0.25

InC

PWSi

nO

C Sin

SoSS

NKI

NKI

Nof

SSE

KIS

KISo

fSe

mI

Sut

Suto

Siin

Sio

Siof UP

UPo

UPo

fU

IU

iof

CI

Ad CD

Off

BSSi

nBS

SoBS

SbBS

Cin

BSC

mBS

Co

BSC

bBS

Cin

BSN

mBS

No

BSN

bAr

inAr

mAr

oAr

b

Prop

ortio

n March

b

0

0.05

0.1

0.15

0.2

0.25

InC

PWSi

nO

C Sin

SoSS

NKI

NKI

Nof

SSE

KIS

KISo

fSe

mI

Sut

Suto

Siin

Sio

Siof UP

UPo

UPo

fU

IU

iof

CI

Ad CD

Off

BSSi

nBS

SoBS

SbBS

Cin

BSC

mBS

Co

BSC

bBS

Cin

BSN

mBS

No

BSN

bAr

inAr

mAr

oAr

b

Prop

ortio

n April

c

0

0.05

0.1

0.15

0.2

0.25

InC

PWSi

nO

C Sin

SoSS

NKI

NKI

Nof

SSE

KIS

KISo

fSe

mI

Sut

Suto

Siin

Sio

Siof UP

UPo

UPo

fU

IU

iof

CI

Ad CD

Off

BSSi

nBS

SoBS

SbBS

Cin

BSC

mBS

Co

BSC

bBS

Cin

BSN

mBS

No

BSN

bAr

inAr

mAr

oAr

b

Nursery areas

Prop

ortio

n May

d

69

Figure 3.5. Proportion of age-0 juveniles that were alive at the end of the simulation (September 1st) in a given nursery area for eggs released in (a) February, (b) March, (c) April and (d) May.

70

Figure 3.6. Connectivity matrix showing transport between spawning and nursery areas for eggs released in February.

InC

PWSi

nO

CS

in SoSS

NKI

NS

SE KIS

Sem

ISu

tSu

toS

Iin Sio

Siof UP

Upo

Upo

fUI

Uio

f CI

Ad

CD Off

BSS

mBS

SoB

SSb

BSC

inBS

Cm

BSC

oBS

CbB

SNin

BSN

mBS

No

BSN

bAr

inA

rm Aro

Arb

InCPWSinOCSinSSNKINSSEKISSEmISutSIinSioUPUPo

Nursery areas

Spa

wni

ng a

reas

0.00-0.05 0.05-0.10 0.10-0.15 0.15-0.20 0.20-0.25 0.25-0.30 0.30-0.35

71

The third most important nursery areas in February were located in the Bering Sea South

Outer domain (BSSo), the Bering Sea Central Middle domain (BSCm), and the Bering Sea

Central Outer domain (BSCo) with 5% of age-0 juveniles (Fig. 3.6). Age-0 juveniles in BSSo

originated from Sin, KIN, Sut, SIo, and UPo (all along the outer edge of the continental shelf in

the GOA). Surviving juveniles were also spawned in SSE, SemI, SIin, and UP, which are located

along the inner edge of the continental shelf in the GOA. BSSb and BSSo showed the same

connectivity routes and transport pattern of particles, ie. these areas were characterized by having

the same spawning areas as the source of individuals (Fig. 3.6). The BSNo nursery area in the

Bering Sea was mainly connected to the UP (20%) spawning area.

Nursery areas BSCm and BSCo contained a primary regions of spawning contributions

which was located on the inner edge of the continental shelf of GOA. Transport proportions to

these areas were highest from the southwest spawning areas and decreased among spawning

regions to the north (i.e. Outer Cook Inlet (OC) toward Shelikof Strait North (SSN), SSE, SemI,

SIin, and UP). In the Bering Sea Central nursery areas, a secondary spawning origin was

associated with the outer edge of the GOA continental shelf, with transport proportions increasing

from Seward inner (Sin) through KIN, KIS, SIo, to UPo (Fig.3.6). Other nursery and spawning

areas had retention values around 5% (e.g. Fig. 3.3a and b), and nursery areas (e.g. SSE and Seml)

with transport values around 5% (Fig.6). .

3.3.4. Spawning in March

The transport of March spawned eggs resembled patterns observed among February

spawned eggs. The highest transport of particles was into SIin in the GOA and BSSb in the BS,

with a transport proportions of 0.15 and 0.11. A group of Bering Sea areas (BSSo, BSCo, BSCm)

all had transport proportions ranging from 5 to 7% (Fig. 3.5b). Transport probabilities increased

by at least 10% in March, relative to February, with values ranging between 45 and 50% (Fig.

3.7). In March, there were two main routes to the SIin area from spawning areas, a primary route

associated with the inner edge of the GOA continental shelf and a secondary route along the outer

edge. SIin also retained particles that were spawned in the region in March. There were two

main nursery areas in the Bering Sea: (1) BSSb and BSSo, which showed the same connectivity

pattern between spawning and nursery areas along the outer edge of the continental shelf (from

Sin to UPo), and secondarily along the inner edge of the continental shelf of the GOA (from SSE

to UP); (2) BSCm and BSCo also contained two routes with origins primarily along the inner edge

72

of the continental shelf. As predicted, contributions to these nursery areas increased with

proximity to the spawning areas. This pattern also occurred along the secondary route to these

Bering Sea nursery areas (i.e., along the outer edge of the shelf) with the exception of UPo where

the transport probability to nursery areas was lower. The small probabilities of retention that

were observed during February in PWSin and OC were intensified during March (Fig. 3.7) to 15-

20% and 10-15%, respectively. The BSNo nursery area was connected to the UP (10-15%)

spawning area, with a level of connectivity that decreased in March compared to that observed

during February.

The nursery area OC was connected to spawning areas PWSin, OC, and Sin (10-15%);

the nursery area Sin was connected with Sin and PWSin; the nursery area SSN was connected to

OC and Sin; and the nursery area KIN was connected to PWSin, all of them showing a transport

probabilities of about 5%. Also, the SSE nursery area was connected to OC (5%) and SemI with

OC (10%) and Sin (5%).

73

Figure 3.7.Connectivity matrix between spawning and nursery areas for the simulation where eggs were released in March.

InC

PW

Sin

OC

Sin

SSN KIN

KIN

of

SSE KIS

Sem

I

Sut

Suto

SIin

Sio

Siof UP

Upo

Upo

f UI

Uio

f CI

Ad CD Off

BSSm

BSSo

BSS

b

BSC

in

BSC

m

BSC

o

BSC

b

BSN

in

BSN

m

BSN

o

BSN

b

Aro

Arb

InC

PWSin

OC

Sin

SSN

KIN

SSE

KIS

SEmI

Sut

SIin

Sio

UP

UPo

Nursery area

Spaw

ning

are

a

0.00-0.05 0.05-0.10 0.10-0.15 0.15-0.20 0.20-0.25 0.25-0.30 0.30-0.35

0.35-0.40 0.40-0.45 0.45-0.50 0.50-0.55

74

3.3.5. Spawning in April

For eggs released in April, some transport patterns to nursery areas differed from previous

spawning months. High transport proportions to SIin in the GOA and BSSb in the BS continued as a

common feature of all months with a transport probabilities of 0.15 and 0.11. In contrast with previous

spawning months, areas associated with Prince William Sound and specifically Outer Cook Inlet (OC)

increased their contributions as nursery areas with a transport proportion of 0.07. Nursery areas in the

Bering Sea such as BSSo, BSCo and BSCm had transport probabilities of 0.09, 0.05 and 0.04 (Fig. 3.5c).

Patterns of retention and connectivity between spawning and nursery areas resembled those

observed for March released eggs, with retention and transport probabilities ranging between 45 and 50%

overall. Contributing spawning areas began as far north as OC, but the connectivity between OC and SIin

was weaker than earlier in the year (Fig. 3.8). The same two routes to the SIin nursery area observed for

February and March released eggs were present during April, with a primary route associated with the inner

edge of the GOA shelf and a secondary route along the outer edge of the GOA shelf. Retention of juveniles

also occurred in all of these areas (Fig.3.8). Two prominent nursery areas occurred in the Bering Sea : (1)

BSSb the primary nursery area, and BSSo, which showed the same pattern of connectivity between

spawning and nursery areas as previously observed, but with additional spawning areas further to the

southwest than in March, and primarily along the outer edge of the continental shelf (from KIN to UPo). A

secondary transport route existed along the inner edge of the continental shelf of the GOA extending from

SSE to UP. For both routes, transport values increased over those observed during March. (2) Areas BSCm

and BSCo also had two connectivity routes that were more contracted to the southwest than in March. As in

previous months, the routes were along the inner edge of the GOA continental shelf, with increasing

transport probabilities from SIin (5%) to UP (25%), and along the outer edge of the GOA continental shelf

originating in Sut (5%) and ending in UPo (25%) (Fig. 3.8). The BSNo nursery area was connected to the

UP (15-20%) spawning area.

A group of nursery areas including OC (15-30%), Sin (5-20%), SSN (5-15%), Sin (5%), and KIN

(5%) all increased transport levels, with the same connectivity routes between spawning and nursery areas

as observed in March. The SSE nursery area was connected with OC (5%) and SemI with OC (5%), Sin

(5%) and SSE (5%).

75

3.3.6. Spawning in May

May released eggs followed the same patterns of retention and connectivity as observed in April,

but the connectivity decreased with values ranging between 40 to 45%. Spawning areas were located

further southwest than those contributing during April (e.g., connectivity from Sin to SIin, Fig. 3.9).

Individuals arriving at SIin followed the inner and outer GOA shelf transport routes that were observed for

individuals released in February, March and April. Fish continued to be retained in these areas (Fig. 3.9).

In the BS the two main nursery areas were present, with intensified connectivity and contributing spawning

areas further southwest than observed for fish spawned in April. Nursery areas BSSb and BSSo maintained

transport routes between spawning and nursery areas, although spawning areas were located further

southwest, along the outer edge of the continental shelf (from KIS to UPo) than in April, and secondarily

along the inner edge of the continental shelf (from Seml to UP). In both cases, transport values or

connectivity increased relative to those observed during April. Nursery areas BSCm and BSCo maintained

the inner and outer edges of the GOA continental shelf primary spawning routes, with increasing transport

from spawning area SIin to UP (5-30%) and Sio to UPo (5-30%) (Fig. 3.9). The BSNo nursery area

maintained its connection to the UP (5%) spawning area, but with reduced connectivity relative to that

observed in April.

Complex interactions in the connectivity of areas inside of the GOA exist in May. OC, PWSin and

InC show significant retention levels. Areas InC and PWSin retained 20% and 30% of individuals spawned

during May. The SSN, KIN, SSE and Sem1 nursery areas show complex connectivity with spawning areas

during this month (see Figure 3.9). The OC nursery area was connected to PWSin (35%) and OC (35%).

The Sin nursery area was connected to PWSin (15%) and Sin (5%). The SSN nursery area was connected

with OC (20%) and KIN was connected with PWSin, OC, and Sin with a connectivity value of 5%. The

SSE nursery area was connected with spawning area InC (5%), OC (10%), Sin (5%), SSN, and KIN at 5%.

The nursery area Seml had two routes of connection, one along the inner edge of the continental shelf

through the SSN (25%) and SSE (15%) spawning areas, and the other along the outer edge of the continental

shelf via KIN (10%) and KIS (5%).

76

Figure 3.8. Connectivity matrix between spawning and nursery areas for the simulation where eggs were released in April.

InC

PWSi

n

OC Sin

SSN

KIN

KIN

of

SSE

KIS

KIS

of

Sem

I

Sut

Sut

o

SIin Sio

Siof UP

Upo

Upo

f UI

Uiof CI

Ad

CD Off

BSS

m

BSSo

BSSb

BSC

in

BSC

m

BSC

o

BSCb

BSN

m

BSN

o

BSN

b

Aro

Arb

InC

PWSin

OC

Sin

SSN

KIN

SSE

KIS

SEmI

Sut

SIin

Sio

UP

UPo

Nursery area

Spaw

ning

are

a

0.00-0.05 0.05-0.10 0.10-0.15 0.15-0.20 0.20-0.25 0.25-0.30 0.30-0.350.35-0.40 0.40-0.45 0.45-0.50 0.50-0.55

77

Figure 3.9 Connectivity matrix between spawning and nursery areas for the simulation where eggs were released in May.

InC

PW

Sin OC

Sin

SS

N

KIN

SS

E

KIS

KIS

of

Sem

I

Sut

Sut

o

SIin Sio

Sio

f

UP

Upo

Upo

f

UI

Uio

f

CI

Ad

CD Off

BS

Sm

BS

So

BS

Sb

BS

Cin

BS

Cm

BS

Co

BS

Cb

BS

No

BS

Nb

Arin

Arm Aro

InC

PWSin

OC

Sin

SSN

KIN

SSE

KIS

SEmI

Sut

SIin

Sio

UP

UPo

Nursery area

Spaw

ning

are

a

0-0.05 0.05-0.1 0.1-0.15 0.15-0.2 0.2-0.25 0.25-0.3 0.3-0.35

0.35-0.4 0.4-0.45 0.45-0.5

78

3.4. Summary of Results Common retention and transport routes from spawning to nursery areas were observed

across all release months. Walleye pollock spawned in the GOA were transported to nursery

areas in the GOA, to the Bering Sea, and to the Aleutians Islands. A constant feature was that the

Shumagin Islands Inner (SIin) area functioned as a retention area (less so in February) and a

nursery destination in the GOA for almost all months. Two primary connectivity routes were

maintained during these simulated months: (1) along the inner edge of the GOA continental shelf,

and (2) a secondary route along the outer edge of the shelf (see Figures 3.10 to 3.13). Nursery

areas in the Bering Sea, BSSb and BSSo, also served as endpoints of routes connecting spawning

areas in the outer and the inner edge of the continental shelf of the GOA to the BS. The

dominance of the outer continental shelf route over the inner route in taking individuals to the

Bering Sea contrasts to the routes taken by individuals to Siin, which was more often along the

inner shelf. Another interesting feature is that the routes of connectivity and the spawning areas

were similar during different spawning months with some variability and with very marked

patterns. Figure 3.10 to 3.13 present summaries of the connectivity and retention in the GOA/AL

and BS.

Transport routes between spawning and nursery areas are summarized by months in Figures 3.10

to 3.13. General patterns observed in the Gulf of Alaska include:

1. The Shumagin Islands inner nursery area, SIin, was connected to spawning areas via a

primary route along the inner edge of the GOA shelf (OC to SIin, ie. through Shelikof Strait),

and a secondary route along the outer edge of the shelf (Sin to SIin). These routes occurred in

all simulated release months (i.e. February to May, Figs. 3.10 to 3.13). The highest

connectivity to the SIin nursery area occurred from spawning occurring in March and April

(Figs. 3.7 and 3.8).

2. In all months the SIin nursery area included retention of eggs that survived to age-0 juveniles.

The highest retention values were observed for individuals spawned in April and May (30-

35%, Figs. 3.8 and 3.9). However, retention was less from the February spawning, the month

when spawning has been observed in this region.

3. As the spawning year progressed, the overall number of retention areas and the retention rates

increased (Figs. 3.10 to 3.13). Prince William Sound and the surrounding areas provided

age-0 walleye pollock retention areas later in the year. Retention in PWS and OC nursery

areas was weak during February (Fig.3.10) and March (Fig.3.11), and increased during April

(Fig. 3.12) and May (Fig. 3.13). Weak retention in Sin occurred in March and was

79

maintained until May. SSE and SemI were secondary nursery areas in terms of the

connectivity between nursery and spawning areas. Other nursery areas appeared in March

through May, such as SSN and KIN.

Two consistent patterns were observed in the retention and transport of walleye pollock early life

stages to Bering Sea nursery areas:

1. The primary nursery area in the Bering Sea, BSSb, and BSSo, shared transport routes

between spawning and nursery areas, where the routes were primarily along the outer edge of

the GOA continental shelf and secondarily along the inner edge of the continental shelf.

2. Transport routes to secondary nursery areas BSCm and BSCo in the Bering Sea were

primarily along the inner edge of the continental shelf (with increasing contributions to the

nursery areas as distance from the spawning area decreased), and secondarily along the outer

edge of the shelf, also with contributions increasing as proximity to spawning areas

decreased. The exception, Upo had a decrease in transport and connectivity with distance

from the spawning area. As the spawning months progressed, the origins of the inner and

outer edge routes moved more to the southwest, with spawning areas closer to the nursery

areas in the BS.

3. Other nursery areas important in the BS included BSCb, BSCo, and BSCm, which were

equally connected to spawning areas along the inner and outer GOA shelf . The

northernmost extent of these spawning areas tended to be restricted to the southwestern areas

in the GOA.

4. In May, nursery areas along the southern Aleutian Islands showed increasing proportions of

surviving juvenile walleye pollock.

80

Figure 3.10. Diagram of connectivity between spawning and nursery areas for individuals spawned in February. The rings in the figure indicate the main nursery areas, rings with arrows represent retention areas. The trajectory arrows represent the routes of connection between spawning and nursery areas, with the width of the arrow proportional to the intensity of connectivity (for values, see Figure 3.6). The names of spawning and nursery areas are indicated

SIin PWSin

OC SemI

SSENursery areas

Connectivity within the GOA in February

Weak Retention

Retention From Spawning areas

Symbols

BSNoBSNm

BSSbBSSo

BSCbBSCoBSCm

Nursery areas

Connectivity between the GOA and the BS in February

Weak Retention


Symbols

81

in Table 3.1, Table 3.2 and Figure 3.1. The color of the arrow indicates the route taken by individuals transport to the nursery area indicated by a circle of the same color.

82

Figure 3.11. Diagram of connectivity between spawning and nursery areas for individuals spawned in March. The rings in the figure indicate the main nursery areas, rings with an arrow represent retention areas. The transport arrows represent the routes of connection between spawning and nursery areas, with the width of the arrow proportional to the intensity of connectivity (for values see Figure 3.7). The names of spawning and nursery areas are indicated

SIin PWSin

OC SemI

SSENursery areas

Weak Retention

Connectivity within the GOA in March

Sin


SymbolsKIN

BSNoBSSo BSCbBSCoBSCm

Nursery areas

Connectivity between the GOA and the BS in March

BSNmBSSb Weak Retention


Symbols

83

in Table 3.1, Table 3.2 and Figure 3.1. . The color of the arrow indicates the route taken by individuals transport to the nursery area indicated by a circle of the same color.

84

Figure 3.12. Diagram of connectivity between spawning and nursery areas for individuals spawned in April. The rings in the figure indicate the main nursery areas, rings with an arrow represent retention areas. The transport arrows represent the routes of connection between spawning and nursery areas, with the width of the arrow proportional to the intensity of

SIin PWSin

OC SemI

SSENursery areas

Weak Retention

Connectivity within the GOA in April

Sin


SymbolsKIN

BSNoBSSoBSSb

BSCbBSCo

Nursery areas

Connectivity between the GOA and the BS in April

Weak Retention


SymbolsUI

85

connectivity (for values see Figure 3.8). The names of spawning and nursery areas are indicated in Table 3.1, Table 3.2 and Figure 3.1. The color of the arrow indicates the route taken by individuals transport to the nursery area indicated by a circle of the same color.

86

Figure 3.13. Diagram of connectivity between spawning and nursery areas for individuals spawned in May. The rings in the figure indicate the main nursery areas, rings with an arrow represent retention areas. The transport arrows represent the routes of connection between spawning and nursery areas, with the width of the arrow proportional to the intensity of

SIin PWSin

OC SemI

SSENursery areas

Weak Retention

Connectivity within the GOA in May

Sin


SymbolsKIN

InC

BSNoBSSo, BSSb BSCb, BSCoNursery areas

Connectivity between the GOA and the BS in May

Weak Retention


SymbolsUI CI

Ad UP

87

connectivity (for values see Figure 3.9). The names of spawning and nursery areas are indicated in Table 3.1, Table 3.2 and Figure 3.1. The color of the arrow indicates the route taken by individuals transport to the nursery area indicated by a circle of the same color.

88

3.5. Discussion The coupled bio-physical IBM was used to examine connections between potential

spawning and nursery areas, and how these patterns changed with changes in the time of

spawning. An unknown portion of the observed variability in spawning area contributions

resulted from not including cumulative mortality when tabulating age-0 juvenile survival, but a

monotonic increase in connectivity with later spawning dates was not observed. Connectivity

between spawning and nursery areas actually declined with later spawning dates in some nursery

areas. But to ensure an equal comparison among spawning dates, cumulative mortality needs to

be included when quantifying connectivity as a function of spawning date. This will be done for

the paper on these results.

Model simulations demonstrated that destination areas for surviving age-0 walleye

pollock juveniles were comparable to observed or suspected nursery areas (Table 3.4). The

modelled nursery areas found in the Semidi Islands (SemI), Sutwik (Sut), and the inner Shumagin

Islands (SIin) corresponded to the observed Shumagin Islands nursery area which extends from

the Semidi Islands to Unimak Pass. The known nursery area north of Kodiak Island

corresponded to modelled nursery areas Outer Cook Inlet (OC) and the inner continental shelf

near Seward (Sin). The modelled nursery area Kodiak Island North (KIN) matched the area

northeast of Kodiak Island where juvenile pollock are found, and the observed nursery area

southwest of Unimak Pass corresponded to the model regions Unimak Pass (UP) and the outer

Unimak Pass (UPo). In contrast to conventional thought, the model region Shumagin Islands

Outer (SIo) was not found to be a regular destination for surviving age-0 walleye pollock in

model simulations.

Walleye pollock spawning locations have shifted in recent years, and other spawning

areas have been recently described. In past years, the majority of pollock spawning in the GOA

was found between mid-March and early May in Shelikof Strait (Fig. 1.1, Kendall et al., 1987;

Schumacher & Kendall 1991). Peak spawning occurred at the beginning of April in the deepest

part of Shelikof Strait. After spawning, eggs and larvae were advected southwest by the Alaska

Coastal Current (ACC) along the Alaska Peninsula and arrived in the Shumagin Islands 8 to 10

weeks later. Another portion of eggs and larvae may have been advected into the Alaskan

Stream, where they could have been lost from the population (Bailey et al. 1999). Walleye

pollock begin to recruit to the harvested population at age-2 and are fully recruited by age 4 or 5

(Bailey et al. 2005). The Shumagin Islands region has traditionally been considered the nursery

89

area that ensures the success of the population in the GOA (Hinckley et al. 1991, Wilson et

al.1996).

In recent years, however, acoustic surveys have observed concentrations of spawning fish

in the Shumagin Islands (Dorn et al., 2005). Little is known about the fate of walleye pollock

spawned in this region, but in stock assessment models, this spawning biomass is assumed to

contribute recruits to the GOA (Dorn et al., 2005). It should be noted that this work indicates

that walleye pollock spawning in February in the Shumagin region mostly were transported to the

Bering Sea; few were retained in this area.

Using model simulation results, destination nursery areas can be predicted for known or

suspected spawning areas. Model results confirmed the connectivity between Shelikof Strait

spawning and the Shumagin nursery region with 40-45% connectivity in March and 45-50% in

April. Connections were also found between SSN and SSE spawning areas and Bering Sea

destination areas such as BSSo (5-10%) and BSSb (5%). These results are relevant as they

confirm that there is a strong connection between the Shelikof and Shumagin areas, and indicate

potential spawning contributions from Shelikof Strait to southern parts of the Bering Sea.

The model also predicted that, of eggs spawned by walleye pollock in the inner parts of

the Shumagin Islands (SIin) during February and March (the observed spawning time in this

region), only 5% were retained in this area, with the rest transported to various nursery areas in

the southern and central Bering Sea (BSSo, 10%; BSSb, 20%; and BSCm, BSCo, BSCb each at

5%). As noted above, this may mean that the Shumagin Islands do not contribute significant

recruits to the GOA, which is what present management assumes.

Prince William Sound ( PWSin), areas north of Kodiak Island ( KIN and SIN), the Sutwik

region (Sut), and the Unimak Pass region (UP and UPo) have also been observed to be spawning

areas. During March and April spawning, PWSin showed a retention rate of 15-25%, with

connectivity to Outer Cook Inlet (OC 10%), the inner shelf at Seward (Sin 5-15%), and Kodiak

Island north (KIN 5%) nursery areas. Spawning-nursery area connectivity from Prince William

Sound spawned walleye pollock may be restricted to the GOA.

Where do juveniles found to the northeast and east of Kodiak Island originate? Our

model indicates that walleye pollock spawned off Seward (Sin), not far from the northern end of

90

Kodiak were only retained at a rate of about 5%. Fish spawned in the North Kodiak area itself

are mostly transported out of the region to the southwest parts of the GOA, to the BS and to the

AL areas in this simulation. A potential caveat to this result is that juvenile fish found in the bays

around northern and eastern Kodiak (Wilson et al., 1996) may have been spawned on the shelf

near Kodiak Island, and they may exhibit behavior and directed swimming not suitable modeled

or simulated by the IBM. The ROMS model, which uses a 10 km grid, does not accurately

resolve flow into and out of coastal bays, and we cannot examine the use of these bays by

juvenile pollock as nursery areas at this time.

Other potential spawning and nursery areas in the GOA could contribute age-0 walleye

pollock juveniles to GOA and BS populations. The Sutwik Island region (Sut) mimicked

retention and transport pattern observed in the Shumagin area (Slin). Individuals from this

region, as well as being retained, were transported to SIin and to the central and southern Bering

Sea. Supporting evidence for contributions from this area is derived from a spatial bioenergetics

model for the western GOA, which indicated that habitat conducive to juvenile walleye pollock

growth was located along the eastern edge of Semidi Bank, in the vicinity of Castle Cape,

Kupreanof Point, and south-west of Sutwik Island (Mazur et al., 2007).

Another important question that may be addressed by this study is whether young pollock

from the GOA are transported to the Bering Sea. Model simulations indicated that the Bering Sea

may be an important nursery area for several spawning grounds in the GOA. This result raises

the possibility of GOA spawning success and transport rates influencing walleye pollock

recruitment in the BS. It is also not known if variability in GOA recruitment can be attributed to

the number of walleye pollock transported to the BS. If surviving GOA juveniles in the BS

develop, grow, and recruit to the BS fishery, then transport of individuals to the BS potentially

influences walleye pollock cohort strength in both the GOA and the BS. The BS pollock stock,

however, is ten times larger than the GOA stock. Also, as juvenile fish in the Bering Sea (age-1’s

and 2’s) tend to be found in the northern half, i.e., west of 170°, any of the juveniles transported

to the BS from the GOA would have to run the gantlet of cannibalistic adult pollock and

increasing numbers of arrowtooth flounder that are found in the southern half of the Bering.

Another possibility, of course, might be that GOA juveniles that settle in the BS may eventually

find their way back to the GOA. Some of the GOA fishery data are suggestive of an influx of

pollock from the Bering Sea (M. Dorn, Alaska Fisheries Science Center, Seattle, WA; pers.

comm.).

91

The bio-physical coupled IBM can potentially contribute to the annual walleye pollock

stock assessments for the BS and GOA. At this time, the model only simulates egg through age-0

juvenile life history stages. There are at least two additional years before individuals are recruited

to the commercial fishery. If GOA spawned walleye pollock are transported to the BS and stay in

the BS, then there are important implications for the assessment and management of Bering Sea

and Gulf of Alaska walleye pollock. If there is an occasional large year class in the Bering Sea

due to an influx of recruits from the GOA, it could distort the stock-recruit relationship, and make

the BS stock seem more productive than it actually is. Additional information on the stock

structure and dynamics of walleye pollock in the northeast Pacific, including potential transport

of walleye pollock early life stages from the GOA to the BS, may decrease error in recruitment

estimates for both regions and help interpret odd patterns in the catch and survey data. Annual

surveys of GOA spawning locations, perhaps from acoustic surveys, would be needed to predict

potential transport to the BS. Ideally, tagging or marking age-0 juveniles would provide

additional information on actual recruit destination and site fidelity of spawning adults.

Validation of model-predicted transport rates would enable the addition of IBM-predicted

survivorship and transport components to the annual stock assessment of BS and GOA walleye

pollock populations.

3.6. Literature Cited

Bailey, K.M., Bond, N.A., Stabeno, P.J., 1999. Anomalous transport of walleye pollock larvae linked to ocean and atmospheric patterns in May 1996. Fisheries Oceanography 8: 264-273. Bailey, K.M. 2000. Shifting control of recruitment of walleye pollock Theragra chalcogramma after a major climatic and ecosystem change. Mar Ecol Prog Ser 198:215-224. Bailey, K. M., L. Ciannelli, N.A. Bond, A. Belgrano, and N. C. Stenseth. 2005. Recruitment of walleye pollock in a physically and biologically complex ecosystem: A new perspective. Prog. Oceanogr. 76:24-42. Beck, M.W., Heck, K.L., Able, K.W. Jr., Childers, D.L., Eggleston, D.B., Gillanders, B.M., Halpern, B., Hays, C.G., Hoshino, H., Minello, T.J., Orth, R.J., Sheridan, P.F., and Weinstein, M.P. 2001. The identification, conservation, and management of estuarine and marine nurseries for fish and invertebrates. BioScience 51: 633-641.

92

Botsford L.W., Hastings A, Gaines S.D. 2001. depdendence of sustainability on the configuration of marine reserves and larval dispersal distance. Ecology letters 4:144-150. Cowen R.K., Lwiza K.M.M., Sponaugle S., Paris C.B., Olson D.B. 2000. Connectivity of marine populations: Open or closed? Science 287:857-859. Cowen R.K., Gawarkiewicz G., Pineda J., Thorrold S.R., Werner F. 2007. Population connectivity in marine systems An overview. Oceanography 20(3):14-21. Curchitser, E.N, D.B. Haidvogel, A.J. Hermann, E.L. Dobbins, T. Powell and A. Kaplan, 2005. Multi-scale modeling of the North Pacific Ocean: Assessment of simulated basin-scale variability (1996-2003). J. Geophys. Res., 110, C11021, doi:101029/2005JC002902. da Silva, A.M., Young, C.C., and S. Levitus. 1994. Atlas of Surface Marine Data 1994. Vol 1. Algorithms and Procedures, NOAA Atlas NESDIS, vol 6, 83 pp., NOAA, Silver Spring, MD. Dorn M, Aydin K, Barbeaux S, Guttormsen M, Megrey B, Spalinger K, Wilkins M. 2005. Assessment of walleye pollock in the Gulf of Alaska In: Appendix B. Stock assessment and fishery evaluation report for the Groundfish Resources of the Gulf of Alaska, North Pacific Fishery Management Council, P.O. Box 103136, Anchorage, AK 99510 Gaines S.D., Caylord B., Gerber L.R., Hastings A., Kinlan B.P. 2007. Connecting places. The ecological consequences of dispersal in the sea. Oceanography 20 (3): 90-99. Grant, W.S. and F.M. Utter.1980 Biochemical genetic variation in walleye pollock, Theragra chalcogramma: population structure in the southeastern Bering Sea and the Gulf of Alaska. Can J Fish Aquat Sci 37:1093–1100. Haidvogel, D.B., H.G. Arango, K. Hedstrom, A.Beckmann, P. Malanotte-Rizzoli, P. and A.F. Shchepetkin. 2000. Model evaluation experiments in the North Atlantic Basin: Simulations in nonlinear terrain-following coordinates, Dynamics Atmosphere Oceans 32: 239-281. Hastings, A. and S. Harrison, 1994. Metapopulation dynamics and genetics. Annual Review of Ecology and Systematics, 25: 167-188. Hermann, A. J., S. Hinckley, B. A. Megrey and P. J. Stabeno. 1996. Interannual variability of the early life history of walleye pollock near Shelikof Strait, as inferred from a spatially explicit, individual-based model.Fish. Oceanogr. 5 (Suppl. 1): 39-57. Hinckley, S., Bailey, K.H., Picquelle, S.J., Shumacher, J.D., Stabeno, P.J., 1991. Transport, distribution, and abundance of larval and juvenile walleye pollock (Theragra chalcogramma) in the western Gulf of Alaska. Can. J. Fish. Aquat. Sci. 48, 91-98. Hinckley S., Hermann A.J., Megrey B.A. 1996. Development of a spatially explicit, individual-based model of marine fish early life history. Marine Ecology progress series139: 47-68. Kendall AW Jr, Clarke ME, Yoklavich MM, Boehlert GW (1987) Distribution, feeding, and growth of larval walleye pollock, Theragra chalcogramma, from Shelikof Strait, Gulf of Alaska. Fish Bull, US 85: 499-521.

93

Kim, S-S, Moon, D-Y and W-S, Yang. 2000. Mitochondrial DNA analysis for stock identification of walleye pollock, Theragra chalcogramma, from the North Pacific Ocean. Bull Nat Fish Res Dev Inst Korea 58:22–27. Levins R. 1969. Some demographic and genetic consequences of environmental heterogeneity for biological control. Bulletin of the Entomological Society of America 15:237-240. Marchesiello, P., McWilliams, J.C., and A. Shchepetkin, 2001: Open boundary conditions for long-term integration of regional ocean models. Ocean Modelling, 3, 1-20. Marchesiello, P., McWilliams, J.C., and A. Shchepetkin, 2003. Equilibrium Structure and Dynamics of the California Current System, J. Phys. Oceanogr., 33, 753-783. Mazur MM, Wilson MT, Dougherty AB, Buchheister A, Beauchamp DA (in press) Temperature and prey quality effects on growth of juvenile walleye pollock Theragra chalcogramma: a spatially explicit bioenergetics approach. J Fish Biol Moore, J. K., D.M. Doney, C. Scott, and K. Lindsay, 2004. Upper ocean ecosystem dynamics and iron cycling in a global three-dimensional model. Global Biogeochemical Cycles 18, [np]. Mulligan, T.J., Chapman, R.W. and B.L., Brown. 1992. Mitochondrial DNA analysis of walleye pollock, Theragra chalcogramma, from the eastern Bering Sea and Shelikof Strait, Gulf of Alaska. Can J Fish Aquat Sci 49:319–326. North et al. in press [Anon2]Olsen, J.B., Merkouris, S.E., Seeb, J.E. 2002. An examination of spatial and temporal genetic variation in walleye pollock (Theragra chalcogramma) using allozyme, mitochondrial DNA, and microsatellite data. Fish Bull 100:752–764. O'Reilly, P.T., Canino, M.F., Bailey. K,M. And P., Bentzen. 2004. Inverse relationship between FST and microsatellite polymorphism in the marine fish, walleye pollock (Theragra chalcogramma): implications for resolving weak population structure. Mol Ecol 13:1799–1814 Pineda J., Hare J., Sponaugle S.U. 2007. Larval transport and dispersal in the coastal ocean and consequences for population connectivity. Oceanography, 20(3): 22-39. Sale P.F., Cowen R.K., Danilowicz B.S., Jones G.P., Kritzer J.P., Lindeman K.C., Planes S., Polunin N.V.C., Russ G.R., Sadovy Y.I., Steneck R.S. 2005. Critical science gaps impede use of no-take fisheries reserves. Trends in Ecology and Evolution 20(2):74-80. Schumacher JD, Kendall AW Jr (1991) Some interactions between young pollock and their environment in the western Gulf of Alaska. Calif Coop Oceanic Fish Invest Rep 32:22-929 Shchepetkin, A.F. and J.C. McWilliams. 2005. The Regional Ocean Modeling System: A split-explicit, free-surface, topography-following coordinate ocean model, Ocean Modeling, 9: 347-404. Shields, G.F. and J.R., Gust. 1995. Lack of geographic structure in mitochondrial DNA sequences of Bering Sea walleye pollock, Theragra chalcogramma. Mol Mar Biol Biotechnol 4:69–82.

94

Werner, F.E., Mackenzie, B.R., Perry, R.I., Lough, R.G., Naimie, C.E., Blanton, B.O., and J.A. Quinlan, Larval trophodynamics, turbulence, and drift on Georges Bank: A sensitivity analysis of cod and haddock. Sci. Mar., 63 (Suppl. 1): 99-115, 2001. Willis T.J., Millar R.B., Babcock R.C., Tolimieri N. 2003. Burdens of evidence and the benefits of marine reserves: Putting Descartes before des horse? Environmental Conservation 30:97-103. Wilson, M.T., Brodeur, R.D., Hinckley, S., 1996. Distribution and abundance of age-0 walleye pollock, Theragra chalcogramma, in the western Gulf of Alaska during September 1990. In: Brodeur RD, Livingston PA, Loughlin TR, Hollowed AB (eds) Ecology of Juvenile Walleye pollock , Theragra chalcogramma, U.S. Dep. Commer., NOAA Tech. Rep. NMFS 126 p 11-24. Wilson, M.T., 2000. Effects of year and region on the abundance and size of age-0 walleye pollock, Theragra chalcogramma, in the western Gulf of Alaska, 1985-1988. Fish. Bull., U.S. 98, 823-834.

95

CONCLUSIONS

Cutting edge methodology has been used to advance our scientific knowledge of walleye

pollock in the North Pacific in this project. Methodological advances include the coupling of a

state-of-the-art hydrodynamic model to our individual-based model of walleye pollock early life

history using a fast and flexible Java interface, and the addition of new features to the IBM to

reflect recent thinking about important recruitment processes for this species. Simulations were

done to test algorithms and parameters, and these resulted in reasonable estimates of growth and

other characteristics of young pollock and their variability, and allowed us to show that we could

replicate known phenomena such as a year of anomalous transport. We now have an application

which was to test the main hypotheses of this project, and indeed which has been used for

recruitment studies of other species in Alaskan waters, such as snow crab (NPRB project 624).

We performed a simulation intended to test whether the coupled models could replicate

the temporal and spatial distribution of several life stages of young pollock for the year 1987, in

which there were multiple surveys of sequential life stages, from early larvae to autumn 0-age

juveniles to compare with model output. The model replicated these patterns of distribution

well, demonstrating the ability of this biophysical model suite to predict the location and timing

of early life stages of walleye pollock in the GOA. The modeled juveniles in this test were found

near the Shumagin Islands, a known nursery area for GOA pollock. This test corroborated the

connection which has been observed between Shelikof Strait spawning and the Shumagin nursery

area (Hinckley et al., 1991). The horizontal movement submodel used for juvenile pollock, based

on a correlated random walk, produced a distribution of early juveniles that agreed with the

survey data, indicating that horizontal movements of 0-age juveniles are important in predicting

their distribution. The model also correctly identified most of the known or hypothesized

juvenile nursery areas in the GOA, such as the Shumagins (as mentioned), the area to the north

and east of Kodiak Island, and around the Semidi Islands. The model corroborated the finding

from parasite studies which found that juvenile walleye pollock found in bays east of Kodiak

Island (Fig. 2.1) did not originate in Shelikof Strait, unless there were anomalous currents (Bailey

et al., 1999). Also, the Bering Sea was predicted to be an important potential nursery area for

GOA spawned pollock, a result that has been hypothesized but had not been tested until this

study. This result has implications for management of walleye pollock in the GOA and the BS,

as the two are currently managed as separate stocks.

96

The results of the connectivity analysis are relevant and important to the ecology of

pollock in the North pacific. There were some important findings from this analysis. As well as

confirming that the coupled bio-physical model could correctly identify most of the known or

suspected juvenile walleye pollock nursery areas in the GOA, the analysis allowed us to examine

possible destinations of young pollock derived from different spawning locations in the GOA,

and to examine the sources of juvenile pollock found in different nursery areas. Model results, as

mentioned above, confirmed the connectivity between Shelikof Strait spawning and the

Shumagin nursery area, but also indicated that some young pollock spawned in Shelikof Strait

may transit to the Bering Sea. The model experiment also indicated that fish spawning in

February in the Shumagin Islands, as has been observed, may not be retained in the GOA. This

spawning aggregation has been suspected of replacing the former large aggregation in Shelikof

Strait in recent years as the most important GOA pollock spawning site in the GOA. The model

results indicate, however, that this area may not represent a significant source of recruits to the

Gulf (recently, the spawning biomass in the Shumagin Islands has only been a small fraction of

that in Shelikof).

Perhaps the most interesting finding from this study is the high percentage of GOA

spawned pollock that may be transported to the Bering Sea from spawning areas widely dispersed

over large portions of the Gulf. Transport of young pollock through Unimak Pass has been

suspected, but for the most part not to the degree indicated by this work. If these juveniles

eventually recruit to the Bering Sea populations, the management of pollock in these two areas as

two separate stocks may be called into question, and recruitment estimates for the two areas may

be confounded. Studies are needed that shed light on this question, and the question of natal

homing of walleye pollock to answer the question of whether juvenile pollock transported from

the GOA to the BS could return to the Gulf to recruit and spawn.

Even if this connection between the GOA and the BS proves to be of lesser magnitude

than indicated by this study, the patterns it has shown may be useful in explaining unusual

patterns in recruitment that are sometimes seen in the data, such as unusually large year classes in

the Bering Sea perhaps due to an influx from the GOA, and perhaps in decreasing the error in

recruitment estimates used in stock assessment models. If the results of this model study could be

validated, IBM-predicted survivorship and transport could be added to the annual stock

assessments of walleye pollock.

97

The immense model-generated “dataset” generated for this work can also be analyzed to

examine qualities that differentiate “survivors” from “nonsurvivors”, ie. young fish that do or do

not make it to nursery areas, and what differentiates them. In recruitment studies it is often said

that, due to the high mortality rates in early life stages and the fact that only about 1% or less of

the initial number of eggs released ever makes it to the age of recruitment, it is the unusual

individual that survives. A unique aspect of individual-based models is that the entire history for

each individual is retained and can be analyzed for patterns of encounter with different

environmental characteristics such as temperature, prey, or currents, or other factors which may

make some individuals more or less likely to be the unusual individual that survives. We hope to

follow this work with a longitudinal analysis of this dataset to discover some of the important

features that allow for survival, and perhaps are key to recruitment success. This type of analysis,

along with this analysis of connectivity between spawning and nursery areas, should allow us to

design model-based indices derived from this coupled model set which could be useful in

forecasting recruitment of walleye pollock.

98

PUBLICATIONS

Parada, C., Hinckley, S., Horne, J., Dorn, M., Hermann, A., Megrey, B. Comparing simulated walleye pollock recruitment indices to data and stock assessment models from the Gulf of Alaska. Accepted. Marine Ecology Progress Series Parada, C., Hinckley,S., Horne, J. and Mazur, M. Empirical corroboration of IBM-predicted walleye pollock (Theragra chalcogramma) spawning-nursery area transport and survival in the Gulf of Alaska. In preparation. Planned submission date: December, 2008. JOURNAL??? Parada, C., Hinckley, S., and Horne, J. Connectivity of Walleye Pollock (Theragra chalcogramma) Spawning and Nursery Areas in the Gulf of Alaska. In preparation. Planned submission date: December, 2008. JOURNAL???

OUTREACH

Web Pages

Ecosystems and Fisheries Oceanography (EcoFOCI) Physical and Biophysical Modeling. Gulf of Alaska and Bering Sea numerical simulation models exploring physical and biological dynamics. http://www.ecofoci.noaa.gov/efoci_models.shtml (In prep) Recruitment Processes Program, Alaska Fisheries Science Center subpage on Ecosystems and ecological modeling subprogram. Will be linked to: http://www.afsc.noaa.gov/RACE/recruitment/default_rp.php

Conference Presentations

Parada, C., S. Hinckley, A.J. Hermann and M. Dorn. “A biophysical model of walleye pollock in the Gulf of Alaska: model to model and model to data comparisons” Advances in Marine Ecosystem Modelling Research Symposium. Plymouth, UK, June 2005.

Hermann, A. J., S. Hinckley, C. Parada, E. Dobbins, C. W. Moore and D. B. Haidvogel, “Immersive visualization: a modern approach for the rapid exploration of coastal ecosystem model dynamics”, Advances in Marine Ecosystem Modelling Research symposium, Jun. 27-29, 2005, Plymouth Marine Laboratory, Plymouth, UK

Hermann, A., M. Dorn and S. Hinckley, “Multi-decadal simulations of circulation and walleye pollock in the Northern Gulf of Alaska”, Fisheries and the Environment (FATE) Science Meeting, Jun. 8-9, 2005, Seattle, WA.

Hinckley, S. and C. Parada. “NPZ & IBM modeling: current Gulf of Alaska biological models, their applications, and future plans”. Presentation to North Pacific Fisheries Management Council. Seattle, WA. Februrary, 2006

Megrey, B., S. Hinckley and C. Parada. “Using IBM’s to compare two hypotheses for the effect of turbulence on feeding rates in larval fishes”. Workshop on advancements in modelling

http://www.ecofoci.noaa.gov/efoci_models.shtml

99

physical-biological interactions in fish early-life history: recommended practices and future directions. Nantes, France. April, 2006.

S. Hinckley. “Future Directions: Integration with observing systems, operational models, monitoring programs, and management recommendations”. Workshop on advancements in modelling physical-biological interactions in fish early-life history: recommended practices and future directions. IFREMER, Nantes, France. April, 2006.

Hermann, A. J., S. Hinckley, C. Parada, E. Dobbins, C. W. Moore and D. B. Haidvogel, “Immersive visualization: a modern approach for the rapid exploration of Eulerian and Individual-Based models”, Workshop on advancements in modelling physical-biological interactions in fish early-life history: recommended practices and future directions (WKAMF), 3-5 April 2006, IFREMER, Nantes, France. Parada, C., S. Hinckley, J. Horne, A.J. Hermann, B. Megrey and M. Dorn. “Hindcasting walleye pollock recruitment and examining pollock stock structure in the Gulf of Alaska using a biophysical model”. Marine Science Symposium, Anchorage, Alaska. January, 2007. Hermann, A. J., S. Hinckley, C. Parada, E. Dobbins, C. W. Moore and D. B. Haidvogel, “Immersive visualization: a modern approach for the rapid exploration of Eulerian and Individual-Based models”, Gordon Conference on Coastal Ocean Modeling, 17-22 June 2007, Colby-Sawyer College, New London NH. Workshop Participation

Hermann, A.J., S. Hinckley. “A brief history of IBM/NPZ/Hydrodynamic modeling in FOCI”. Joint Centre de Recherche Halieutique (CRH) seminar, June, 2005, Sete, France.

Hermann, A. J. “Immersive visualization techniques for the exploration of hydrodynamic and biological model output”, Fisheries Oceanographic Modeling Workshop, Centre de Recherche Halieutique (CRH), , June 24, 2005, Sete, France.

Hinckley, S. “NPZ and individual-based models in the Bering Sea/Aleutian Islands”. Workshop on evaluation of ocean circulation models for the Bering Sea and Aleutian Islands Region”. NPRB Project 402. February, 2005.

Workshop on advancements in modelling physical-biological interactions in fish early-life history: recommended practices and future directions (WKAMF). Nantes, France. April, 2006.

Presentations in Schools

Hermann, A. J., E. N. Curchitser, D. B. Haidvogel, E. L. Dobbins, S. Hinckley and C. W. Moore, "Immersive visualization of the circulation and biology of the Northeast Pacific using spatially nested models", Physical Oceanography Seminar, College of Oceanic and Atmospheric Sciences, Oregon State University, Feb 16, 2005, Corvallis, OR USA

100

Parada, C. Introduction to coupled biophysical models with IBMs. University of Concepcion summer school, December, 2005. Hermann, A. J.,"Immersive 3D visualization of Lagrangian and Eulerian phenomena in oceanography”, University of Washington Physical Oceanography seminar, May 31, 2006, University of Washington, Seattle, WA USA. Hermann, A. J. “3D Astronomy and Oceanography”, Whittier Elementary Science Fair, March 28, 2006, Seattle, WA USA. Parada, C. Individual-based models in fisheries oceanography. University of Concepcion. December, 2007. Hinckley, S. “Applied individual-based models in fisheries oceanography”. Ecological modeling class, University of Washington, Seattle. May, 2007 and May, 2008

ACKNOWLEDGEMENTS

Be brief and include other funding agencies if appropriate.

LITERATURE CITED

List only those references not listed in the chapters, i.e., those used in the Introduction or Conclusions.

101

PROJECT SYNOPSIS Introduction: Walleye pollock – backbone of an ecosystem a fishery, and many people’s livelihood Walleye pollock (Theragra chalcogramma) in Alaska supports one of the largest fisheries in the world, as well as being a pivotal species in the ecosystems in the Gulf of Alaska. Walleye pollock spawn in many different locations in the Gulf of Alaska (GOA), and the nursery areas for juvenile fish (less than 1 year old) are found in many other locations. At the present time, little is known about where the young fish from each spawning area end up, or to put it another way, where the juveniles found in a particular location were spawned. In other words the connectivity or pathways between spawning and nursery grounds is not well understood. Female walleye pollock spawn millions of eggs, but 99% of these die before the end of their first year, due to some combination of three factors: lack of food, high levels of predation, and transport by ocean currents out of their preferred habitat. The number of fish left to support both the ecosystem and the fisheries (“recruits”) varies widely among years. This variability can put stress on species in the ecosystem that depend on juvenile or adult pollock for food such as Steller sea lions, fur seals, many groundfish and bird species, among others, and also on the pollock fishery itself. Why we did it: Managers of Alaskan ecosystems and fisheries need to understand how the number of recruits varies with the number of spawners and with variability in the physical and biological environment. They must have an understanding of whether spawning and juvenile nursery areas for walleye pollock represent separate populations or a single Gulf of Alaska or Bering Sea stock, to correctly parameterize the stock assessment models, set the management regime and ultimately, correctly establish catch quotas. Only when spawning-nursery area connections are understood, can the reasons for variations in “recruitment” both for near-term (1-3 years) and long-term (decades) be clarified. This is important to manage removals by the fishery in a responsible way, leaving enough pollock for other “mouths” in the ecosystem. Projections of how walleye pollock recruitment may be affected as climate changes will also help in long-term management of this species. How we did it: We developed a set of coupled physical and biological simulation models to track the trajectories through the ocean, the feeding and growth, predation and survival of young pollock in the first year of their life. The physical model gives us a 3D “movie” of the currents, and temperature and salinity patterns as they change over time. In the biological model, young fish are moved around by the currents, and physical factors at different locations affect processes such as encounter with food, metabolic and development rates, growth and mortality. With these models, we can do “experiments” designed to tell us what combination of factors affect survival in any year, and how this may change as the environment changes. In the main experiment done for this project, we released virtual walleye pollock eggs over the whole spawning range of the species in the GOA, and followed them until the fall of their first year when they were juveniles. This way we could track the links between spawning time and location, and nursery areas, and the conditions along the way. What we discovered: We verified with the model that in the GOA many young pollock use the Shumagin Island region as a nursery area. The model tells us that these fish could be spawned along the inner part of the continental shelf of the GOA, in areas such as Outer Cook Inlet, Shelikof Strait and the Semidi Islands, as well as in the Shumagin region itself. An interesting

102

result is that many young pollock may be born in the GOA, but end up in the Bering Sea, especially if they were spawned on the outer edges of the continental shelf or slope; or in the Shumagin region. This is important, as it raises two questions: (1) are the GOA and Bering Sea walleye pollock populations really separate, or is recruitment in the Bering Sea affected by spawning in the Gulf?, and (2) the Shumagin Island region has recently found to contain many spawning pollock – but is this a self-sustaining population, or are all the young fish produced in this region “lost” to the Gulf? What percentage of those spawned here remain? These are critical factors for managers to understand. What’s next: Our coupled model set may be a useful tool to forecast recruitment in such a way that includes mechanisms, i.e. the physical and biological factors that are most important in controlling recruitment variability. To do this, we need to understand what differentiates “survivors” (those that eventually will recruit) and “non survivors”. The output of the model experiment done for this project should be the foundation to a subsequent effort to telling us this. We can track each individual to understand why it survived or didn’t – too little food? Too much cold water? Was it carried out to sea? Once these factors are better understood, we can design model-derived indices that may allow us to forecast future recruitment. Outreach:

• Nine presentations at scientific conferences, including the Advances in Marine Ecosystem Modelling Research Symposium, the Fisheries and the Environment Science Meeting, the Advancements in Modelling Physical-Biological Interactions in Fish Early-Life History Conference, the Alaska Marine Science Symposium, and the Gordon Conference on Coastal Ocean Modeling.

• One presentation to the North Pacific Fisheries Management Council. • Participation in three workshops: (1) the Fisheries oceanographic modeling workshop,

Centre de Recherche Halieutique (CRH), (2) the Workshop on evaluation of ocean circulation models for the Bering Sea and Aleutian Islands Region (NPRB Project 402), and (3) the Workshop on advancements in modelling physical-biological interactions in fish early-life history: recommended practices and future directions.

• Several school presentations, including Whittier Elementary School, the University of Washington Physical Oceanography Department, the UW School of Fisheries and Aquatic Sciences and the UW Quantitative Ecology and Resource Management program, at Oregon State University Physical Oceanography Department, and at the University of Concepcion (Chile).

• The EcoFOCI website (www.ecofoci.noaa.gov) posted some of the presentations of the walleye pollock project results and work on a specific website for this work is in progress.

• Two courses on individual-based modeling for graduate students were organized and conducted, for training and dissemination of the biophysical coupling techniques with emphasis on the Gulf of Alaska walleye pollock case study.

The Big Picture: Biophysical modeling is an important tool to help us understand how to better manage fish stocks for both the fisheries that depend on them and other creatures in the environment that also depend on them. This is because we can perform “experiments” with models in ways that tell us about mechanisms – ie. how the systems actually work, and how climate change may affect these systems. In this study, we discovered important relationships between walleye pollock spawning in different places within the Gulf of Alaska, and between the

103

GOA and the Bering Sea juvenile nursery areas. This should help managers effectively manage this species. NPRB Research Interest: The overall mission of NPRB is to support research “designed to address pressing fishery management or marine ecosystem information needs.” This research will add to the ability of managers to protect the walleye pollock fishery, and the marine ecosystems in the GOA and the BS for which pollock is a pivotal species. It will add to the knowledge required to use this coupled biophysical model to forecast recruitment levels for pollock in both the short and the long term. We have developed a tool that can be used with other species in other areas as well.

north pacific research board project final reportdepts.washington.edu/fish437/resources/week...

Documents